Map-based methods for autonomous racing estimate the vehicle's location, which is used to follow ... more Map-based methods for autonomous racing estimate the vehicle's location, which is used to follow a highlevel plan. While map-based optimisation methods demonstrate high-performance results, they are limited by requiring a map of the environment. In contrast, mapless methods can operate in unmapped contexts since they directly process raw sensor data (often LiDAR) to calculate commands. However, a major limitation in mapless methods is poor performance due to a lack of optimisation. In response, we propose the local map framework that uses easily extractable, low-level features to build local maps of the visible region that form the input to optimisation-based controllers. Our local map generation extracts the visible racetrack boundaries and calculates a centreline and track widths used for planning. We evaluate our method for simulated F1Tenth autonomous racing using a two-stage trajectory optimisation and tracking strategy and a model predictive controller. Our method achieves lap times that are 8.8% faster than the Follow-The-Gap method and 3.22% faster than end-to-end neural networks due to the optimisation resulting in a faster speed profile. The local map planner is 3.28% slower than global methods that have access to an entire map of the track that can be used for planning. Critically, our approach enables high-speed autonomous racing on unmapped tracks, achieving performance similar to global methods without requiring a track map.
The F1TENTH autonomous driving platform, consisting of 1:10-scale remote-controlled cars, has evo... more The F1TENTH autonomous driving platform, consisting of 1:10-scale remote-controlled cars, has evolved into a well-established education and research platform. The many publications and real-world competitions span many domains, from classical path planning to novel learning-based algorithms. Consequently, the field is wide and disjointed, hindering direct comparison of developed methods and making it difficult to assess the state-of-the-art. Therefore, we aim to unify the field by surveying current approaches, describing common methods, and providing benchmark results to facilitate clear comparisons and establish a baseline for future work. This research aims to survey past and current work with F1TENTH vehicles in the classical and learning categories and explain the different solution approaches. We describe particle filter localisation, trajectory optimisation and tracking, model predictive contouring control, follow-the-gap, and end-to-end reinforcement learning. We provide an open-source evaluation of benchmark methods and investigate overlooked factors of control frequency and localisation accuracy for classical methods as well as reward signal and training map for learning methods. The evaluation shows that the optimisation and tracking method achieves the fastest lap times, followed by the online planning approach. Finally, our work identifies and outlines the relevant research aspects to help motivate future work in the F1TENTH domain.
2021 20th International Conference on Advanced Robotics (ICAR), Dec 6, 2021
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
The classical method of autonomous racing uses real-time localisation to follow a precalculated o... more The classical method of autonomous racing uses real-time localisation to follow a precalculated optimal trajectory. In contrast, end-to-end deep reinforcement learning (DRL) can train agents to race using only raw LiDAR scans. While classical methods prioritise optimization for high-performance racing, DRL approaches have focused on low-performance contexts with little consideration of the speed profile. This work addresses the problem of using end-to-end DRL agents for highspeed autonomous racing. We present trajectory-aided learning (TAL) that trains DRL agents for high-performance racing by incorporating the optimal trajectory (racing line) into the learning formulation. Our method is evaluated using the TD3 algorithm on four maps in the open-source F1Tenth simulator. The results demonstrate that our method achieves a significantly higher lap completion rate at high speeds compared to the baseline. This is due to TAL training the agent to select a feasible speed profile of slowing down in the corners and roughly tracking the optimal trajectory.
Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for auton... more Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for autonomous motion planning. The application of RL to a specific problem is dependent on a reward signal to quantify how good or bad a certain action is. This paper addresses the problem of reward signal design for robotic control in the context of local planning for autonomous racing. We aim to design reward signals that are able to perform well in multiple, competing, continuous metrics. Three different methodologies of position-based, velocity-based, and action-based rewards are considered and evaluated in the context of F1/10th racing. A novel method of rewarding the agent on its state relative to an optimal trajectory is presented. Agents are trained and tested in simulation and the behaviors generated by the reward signals are compared to each other on the basis of average lap time and completion rate. The results indicate that a reward based on the distance and velocity relative to a minimum curvature trajectory produces the fastest lap times.
2021 20th International Conference on Advanced Robotics (ICAR), Dec 6, 2021
Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for auton... more Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for autonomous motion planning. The application of RL to a specific problem is dependent on a reward signal to quantify how good or bad a certain action is. This paper addresses the problem of reward signal design for robotic control in the context of local planning for autonomous racing. We aim to design reward signals that are able to perform well in multiple, competing, continuous metrics. Three different methodologies of position-based, velocity-based, and action-based rewards are considered and evaluated in the context of F1/10th racing. A novel method of rewarding the agent on its state relative to an optimal trajectory is presented. Agents are trained and tested in simulation and the behaviors generated by the reward signals are compared to each other on the basis of average lap time and completion rate. The results indicate that a reward based on the distance and velocity relative to a minimum curvature trajectory produces the fastest lap times.
2020 International SAUPEC/RobMech/PRASA Conference, 2020
The task of a quadrotor carrying an unknown suspended payload is difficult, as the payload has a ... more The task of a quadrotor carrying an unknown suspended payload is difficult, as the payload has a considerable effect on the flight dynamics of the vehicle, the payload parameters are unknown and the payload state is not available for measurement. Linear Quadratic Gaussian control is considered to address the problem. The payload parameters are first estimated using Recursive Least Squares to estimate the payload mass and estimating the cable length from the oscillating frequency obtained with the Fast-Fourier Transform. These estimated parameters are used to design the Linear Quadratic Gaussian controller to simultaneously estimate the swing angle of the payload and apply full-state feedback control. These techniques are found to reduce the oscillations caused by the payload, even in the presence of sensor noise and external disturbances, which practical systems suffer from.
Deep reinforcement learning (DRL) is a promising method to learn control policies for robots only... more Deep reinforcement learning (DRL) is a promising method to learn control policies for robots only from demonstration and experience. To cover the whole dynamic behaviour of the robot, DRL training is an active exploration process typically performed in simulation environments. Although this simulation training is cheap and fast, applying DRL algorithms to real-world settings is difficult. If agents are trained until they perform safely in simulation, transferring them to physical systems is difficult due to the sim-to-real gap caused by the difference between the simulation dynamics and the physical robot. In this paper, we present a method of online training a DRL agent to drive autonomously on a physical vehicle by using a model-based safety supervisor. Our solution uses a supervisory system to check if the action selected by the agent is safe or unsafe and ensure that a safe action is always implemented on the vehicle. With this, we can bypass the sim-toreal problem while training the DRL algorithm safely, quickly, and efficiently. We compare our method with conventional learning in simulation and on a physical vehicle. We provide a variety of real-world experiments where we train online a smallscale vehicle to drive autonomously with no prior simulation training. The evaluation results show that our method trains agents with improved sample efficiency while never crashing, and the trained agents demonstrate better driving performance than those trained in simulation.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The paper describes the development and simulation results of a full ADCS subsystem for the deOrb... more The paper describes the development and simulation results of a full ADCS subsystem for the deOrbitSail drag sail mission. The deOrbitSail satellite was developed as part of an European FP7 collaboration research project. The satellite was launched and commissioning started on 10th July 2015. Various new actuators and sensors designed for this mission will be presented. The deOrbitSail satellite is a 3U CubeSat to deploy a 4 by 4 m drag sail from an initial 650 km circular polar low earth orbit. With an active attitude control system it will be shown that by maximising the drag force, the expected de-orbiting period from the initial altitude will be less than 50 days. A future application of this technology will be the use of small drag sails as low-cost devices to de-orbit LEO satellites, when they have reached their end of life, without having to use expensive propulsion systems. Simulation and Hardware-in-Loop experiments proved the feasibility of the proposed attitude control system. A magnetic-only control approach using a Y-Thomson spin, is used to detumble the 3U Cubesat with stowed sail and subsequently to 3-axis stabilise the satellite to be ready for the final deployment phase. Minituarised torquer rods, a nano-sized momentum wheel, attitude sensor hardware (magnetometer, sun, earth) developed for this phase will be presented. The final phase will be to deploy and 3-axis stabilise the drag sail normal to the satellite's velocity vector, using a combined Y-momentum wheel and magnetic controller. The design and performance improvements when using a 2-axis translation stage to adjust the sail centre-of-pressure to satellite centre-of-mass offset, will also be discussed, although for launch risk reasons this stage was not included in the final flight configuration. To accurately determine the drag sail's attitude during the sunlit part of the orbit, an accurate wide field of view dual sensor to measure both the sun and nadir vector direction was developed for this mission. The calibration results for this new Cubesat sensor (CubeSense), will also be presented.
Fused deposition modelling 3D printers are a promising method to manufacture cold gas thruster no... more Fused deposition modelling 3D printers are a promising method to manufacture cold gas thruster nozzles for air bearing test facilities and similar robotics test platforms. This paper investigates the performance and feasibility of 3D printed cold gas thruster nozzles. Performance parameters included thrust and specific impulse. Various slicer settings, nozzle profiles and a range of thrust outputs were compared. It is concluded that the nozzles perform reasonably well when designed for thrusts in the range of 50 to 250 mN, where they achieve 60 to 90 % of their theoretical specific impulse. For nozzles which should output more than 250 mN of thrust it was found that friction causes significant losses, and other manufacturing technologies will likely perform better.
This paper addresses the problem of a multirotor carrying an unknown suspended payload, allowed t... more This paper addresses the problem of a multirotor carrying an unknown suspended payload, allowed to swing in one axis. The payload is unknown in the sense that its parameters, such as the payload mass and cable length, are unknown and its state, the swing angles, are not available for measurement. The suspended payload alters the flight dynamics of the vehicle considerably and the flight controllers need to minimize this effect. A robust model reference adaptive control technique is designed and implemented to minimize the effect of the suspended payload on the vehicle while allowing the controller to adapt to account for the unknown payload. The controller is modified to reject external disturbances and to be robust in the presence of sensor noise and drift. In simulation, this technique proves to dampen the oscillations caused by the payload. A quadrotor was built to practically demonstrate the effectiveness of the controller. The PX4 flight control stack is considered for the firmware of the vehicle. The model reference adaptive controller was implemented and succeeded to dampen the oscillations caused by the payload in a practical flight.
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
Recent successes in the deployment of sails in space have reduced the risk associated with solar ... more Recent successes in the deployment of sails in space have reduced the risk associated with solar sailing missions. The attitude control requirements for a solar sailing mission is low with only slow attitude maneuvers needed to maintain a stable attitude and produce a required solar thrust. Future science missions will require large attitude maneuvers with a fully deployed sail. This article investigates the current options for attitude control on solar sails and proposes a gyrocontrolled solar sailing. This concept uses a spinning solar sail to construct a control moment gyroscope capable to produce large torques. Steering laws for performing attitude maneuvers and simulation results are presented which demonstrates the capabilities of such a solution.
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Map-based methods for autonomous racing estimate the vehicle's location, which is used to follow ... more Map-based methods for autonomous racing estimate the vehicle's location, which is used to follow a highlevel plan. While map-based optimisation methods demonstrate high-performance results, they are limited by requiring a map of the environment. In contrast, mapless methods can operate in unmapped contexts since they directly process raw sensor data (often LiDAR) to calculate commands. However, a major limitation in mapless methods is poor performance due to a lack of optimisation. In response, we propose the local map framework that uses easily extractable, low-level features to build local maps of the visible region that form the input to optimisation-based controllers. Our local map generation extracts the visible racetrack boundaries and calculates a centreline and track widths used for planning. We evaluate our method for simulated F1Tenth autonomous racing using a two-stage trajectory optimisation and tracking strategy and a model predictive controller. Our method achieves lap times that are 8.8% faster than the Follow-The-Gap method and 3.22% faster than end-to-end neural networks due to the optimisation resulting in a faster speed profile. The local map planner is 3.28% slower than global methods that have access to an entire map of the track that can be used for planning. Critically, our approach enables high-speed autonomous racing on unmapped tracks, achieving performance similar to global methods without requiring a track map.
The F1TENTH autonomous driving platform, consisting of 1:10-scale remote-controlled cars, has evo... more The F1TENTH autonomous driving platform, consisting of 1:10-scale remote-controlled cars, has evolved into a well-established education and research platform. The many publications and real-world competitions span many domains, from classical path planning to novel learning-based algorithms. Consequently, the field is wide and disjointed, hindering direct comparison of developed methods and making it difficult to assess the state-of-the-art. Therefore, we aim to unify the field by surveying current approaches, describing common methods, and providing benchmark results to facilitate clear comparisons and establish a baseline for future work. This research aims to survey past and current work with F1TENTH vehicles in the classical and learning categories and explain the different solution approaches. We describe particle filter localisation, trajectory optimisation and tracking, model predictive contouring control, follow-the-gap, and end-to-end reinforcement learning. We provide an open-source evaluation of benchmark methods and investigate overlooked factors of control frequency and localisation accuracy for classical methods as well as reward signal and training map for learning methods. The evaluation shows that the optimisation and tracking method achieves the fastest lap times, followed by the online planning approach. Finally, our work identifies and outlines the relevant research aspects to help motivate future work in the F1TENTH domain.
2021 20th International Conference on Advanced Robotics (ICAR), Dec 6, 2021
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
The classical method of autonomous racing uses real-time localisation to follow a precalculated o... more The classical method of autonomous racing uses real-time localisation to follow a precalculated optimal trajectory. In contrast, end-to-end deep reinforcement learning (DRL) can train agents to race using only raw LiDAR scans. While classical methods prioritise optimization for high-performance racing, DRL approaches have focused on low-performance contexts with little consideration of the speed profile. This work addresses the problem of using end-to-end DRL agents for highspeed autonomous racing. We present trajectory-aided learning (TAL) that trains DRL agents for high-performance racing by incorporating the optimal trajectory (racing line) into the learning formulation. Our method is evaluated using the TD3 algorithm on four maps in the open-source F1Tenth simulator. The results demonstrate that our method achieves a significantly higher lap completion rate at high speeds compared to the baseline. This is due to TAL training the agent to select a feasible speed profile of slowing down in the corners and roughly tracking the optimal trajectory.
Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for auton... more Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for autonomous motion planning. The application of RL to a specific problem is dependent on a reward signal to quantify how good or bad a certain action is. This paper addresses the problem of reward signal design for robotic control in the context of local planning for autonomous racing. We aim to design reward signals that are able to perform well in multiple, competing, continuous metrics. Three different methodologies of position-based, velocity-based, and action-based rewards are considered and evaluated in the context of F1/10th racing. A novel method of rewarding the agent on its state relative to an optimal trajectory is presented. Agents are trained and tested in simulation and the behaviors generated by the reward signals are compared to each other on the basis of average lap time and completion rate. The results indicate that a reward based on the distance and velocity relative to a minimum curvature trajectory produces the fastest lap times.
2021 20th International Conference on Advanced Robotics (ICAR), Dec 6, 2021
Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for auton... more Reinforcement learning (RL) has shown to be a valuable tool in training neural networks for autonomous motion planning. The application of RL to a specific problem is dependent on a reward signal to quantify how good or bad a certain action is. This paper addresses the problem of reward signal design for robotic control in the context of local planning for autonomous racing. We aim to design reward signals that are able to perform well in multiple, competing, continuous metrics. Three different methodologies of position-based, velocity-based, and action-based rewards are considered and evaluated in the context of F1/10th racing. A novel method of rewarding the agent on its state relative to an optimal trajectory is presented. Agents are trained and tested in simulation and the behaviors generated by the reward signals are compared to each other on the basis of average lap time and completion rate. The results indicate that a reward based on the distance and velocity relative to a minimum curvature trajectory produces the fastest lap times.
2020 International SAUPEC/RobMech/PRASA Conference, 2020
The task of a quadrotor carrying an unknown suspended payload is difficult, as the payload has a ... more The task of a quadrotor carrying an unknown suspended payload is difficult, as the payload has a considerable effect on the flight dynamics of the vehicle, the payload parameters are unknown and the payload state is not available for measurement. Linear Quadratic Gaussian control is considered to address the problem. The payload parameters are first estimated using Recursive Least Squares to estimate the payload mass and estimating the cable length from the oscillating frequency obtained with the Fast-Fourier Transform. These estimated parameters are used to design the Linear Quadratic Gaussian controller to simultaneously estimate the swing angle of the payload and apply full-state feedback control. These techniques are found to reduce the oscillations caused by the payload, even in the presence of sensor noise and external disturbances, which practical systems suffer from.
Deep reinforcement learning (DRL) is a promising method to learn control policies for robots only... more Deep reinforcement learning (DRL) is a promising method to learn control policies for robots only from demonstration and experience. To cover the whole dynamic behaviour of the robot, DRL training is an active exploration process typically performed in simulation environments. Although this simulation training is cheap and fast, applying DRL algorithms to real-world settings is difficult. If agents are trained until they perform safely in simulation, transferring them to physical systems is difficult due to the sim-to-real gap caused by the difference between the simulation dynamics and the physical robot. In this paper, we present a method of online training a DRL agent to drive autonomously on a physical vehicle by using a model-based safety supervisor. Our solution uses a supervisory system to check if the action selected by the agent is safe or unsafe and ensure that a safe action is always implemented on the vehicle. With this, we can bypass the sim-toreal problem while training the DRL algorithm safely, quickly, and efficiently. We compare our method with conventional learning in simulation and on a physical vehicle. We provide a variety of real-world experiments where we train online a smallscale vehicle to drive autonomously with no prior simulation training. The evaluation results show that our method trains agents with improved sample efficiency while never crashing, and the trained agents demonstrate better driving performance than those trained in simulation.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The paper describes the development and simulation results of a full ADCS subsystem for the deOrb... more The paper describes the development and simulation results of a full ADCS subsystem for the deOrbitSail drag sail mission. The deOrbitSail satellite was developed as part of an European FP7 collaboration research project. The satellite was launched and commissioning started on 10th July 2015. Various new actuators and sensors designed for this mission will be presented. The deOrbitSail satellite is a 3U CubeSat to deploy a 4 by 4 m drag sail from an initial 650 km circular polar low earth orbit. With an active attitude control system it will be shown that by maximising the drag force, the expected de-orbiting period from the initial altitude will be less than 50 days. A future application of this technology will be the use of small drag sails as low-cost devices to de-orbit LEO satellites, when they have reached their end of life, without having to use expensive propulsion systems. Simulation and Hardware-in-Loop experiments proved the feasibility of the proposed attitude control system. A magnetic-only control approach using a Y-Thomson spin, is used to detumble the 3U Cubesat with stowed sail and subsequently to 3-axis stabilise the satellite to be ready for the final deployment phase. Minituarised torquer rods, a nano-sized momentum wheel, attitude sensor hardware (magnetometer, sun, earth) developed for this phase will be presented. The final phase will be to deploy and 3-axis stabilise the drag sail normal to the satellite's velocity vector, using a combined Y-momentum wheel and magnetic controller. The design and performance improvements when using a 2-axis translation stage to adjust the sail centre-of-pressure to satellite centre-of-mass offset, will also be discussed, although for launch risk reasons this stage was not included in the final flight configuration. To accurately determine the drag sail's attitude during the sunlit part of the orbit, an accurate wide field of view dual sensor to measure both the sun and nadir vector direction was developed for this mission. The calibration results for this new Cubesat sensor (CubeSense), will also be presented.
Fused deposition modelling 3D printers are a promising method to manufacture cold gas thruster no... more Fused deposition modelling 3D printers are a promising method to manufacture cold gas thruster nozzles for air bearing test facilities and similar robotics test platforms. This paper investigates the performance and feasibility of 3D printed cold gas thruster nozzles. Performance parameters included thrust and specific impulse. Various slicer settings, nozzle profiles and a range of thrust outputs were compared. It is concluded that the nozzles perform reasonably well when designed for thrusts in the range of 50 to 250 mN, where they achieve 60 to 90 % of their theoretical specific impulse. For nozzles which should output more than 250 mN of thrust it was found that friction causes significant losses, and other manufacturing technologies will likely perform better.
This paper addresses the problem of a multirotor carrying an unknown suspended payload, allowed t... more This paper addresses the problem of a multirotor carrying an unknown suspended payload, allowed to swing in one axis. The payload is unknown in the sense that its parameters, such as the payload mass and cable length, are unknown and its state, the swing angles, are not available for measurement. The suspended payload alters the flight dynamics of the vehicle considerably and the flight controllers need to minimize this effect. A robust model reference adaptive control technique is designed and implemented to minimize the effect of the suspended payload on the vehicle while allowing the controller to adapt to account for the unknown payload. The controller is modified to reject external disturbances and to be robust in the presence of sensor noise and drift. In simulation, this technique proves to dampen the oscillations caused by the payload. A quadrotor was built to practically demonstrate the effectiveness of the controller. The PX4 flight control stack is considered for the firmware of the vehicle. The model reference adaptive controller was implemented and succeeded to dampen the oscillations caused by the payload in a practical flight.
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
Recent successes in the deployment of sails in space have reduced the risk associated with solar ... more Recent successes in the deployment of sails in space have reduced the risk associated with solar sailing missions. The attitude control requirements for a solar sailing mission is low with only slow attitude maneuvers needed to maintain a stable attitude and produce a required solar thrust. Future science missions will require large attitude maneuvers with a fully deployed sail. This article investigates the current options for attitude control on solar sails and proposes a gyrocontrolled solar sailing. This concept uses a spinning solar sail to construct a control moment gyroscope capable to produce large torques. Steering laws for performing attitude maneuvers and simulation results are presented which demonstrates the capabilities of such a solution.
The problem of autonomous racing is to navigate through a race course as quickly as possible whil... more The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10 th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Uploads
Papers by Willem Jordaan