Learning Fast Bipedal Locomotion Master’s Thesis
نویسندگان
چکیده
In this thesis, I present a method to optimize the clock-driven, periodical walking pattern of a humanoid robot for forward speed using metaheuristics. I start from a hand-tuned open-loop gait and enhance it with two feedback control mechanisms. First, I employ a P-controller to regulate the foot angle to reduce angular velocity of the robot’s body. The angular velocity is measured by a gyroscope. Second, I propose a phase resetting mechanism that resets the step cycle at the moment of foot contact. In simulated experiments, I demonstrate that feedback control is essential to achieve fast and robust locomotion. To derive an adequate metaheuristic for optimizing the gait, I compare a Downhill Simplex search, a policy gradient reinforcement learning approach (PGRL), and Particle Swarm Optimization. I demonstrate in simulated experiments that PGRL is the most effective algorithm for solving the optimization problem considered in this thesis. I extend the PGRL algorithm by an adaptive step size and a sequential sampling procedure in order to reduce the number of evaluations of the fitness function. Furthermore, I prove that my extensions to the PGRL increase the performance of the algorithm significantly in terms of the quality of the found solutions. I used the extended PGRL algorithm to optimize the gait of a real robot. After optimizing the open-loop trajectory generation parameters and the feedback parameters of the gait, the robot can walk at a speed of 34cm/s. This corresponds to a gain of over 50% compared to the former top speed achieved using a hand-tuned gait.
منابع مشابه
Reinforcement Learning Inspired Disturbance Rejection and Nao Bipedal Locomotion
Competitive bipedal soccer playing robots need to move fast and react quickly to changes in direction while staying upright. This paper describes the application of reinforcement learning to stabilise a flat-footed humanoid robot. An optimal control policy is learned using a physics simulator. The learned policy is supported theoretically and interpreted on a real robot as a linearised continuo...
متن کاملThesis Proposal Learning optimal policies for compliant gaits and their implementation on robot hardware
Bipedal animals exhibit a diverse range of gaits and gait transitions, which can robustly travel over terrains of varying grade, roughness, and compliance. Bipedal robots should be capable of the same. Despite these clear goals, stateof-the-art humanoid robots have not yet demonstrated locomotion behaviors that are as robust or varied as those of humans and animals. Current modelbased controlle...
متن کاملExploiting Human Motor Skills for Training Bipedal Robots Undergraduate Honors Thesis
Although machine learning, reinforcement learning, and learning from demonstration have improved the rate and accuracy at which robots can gain intelligence from humans, they haven’t reached the rapid rate at which humans are able to acquire new knowledge. Many systems that exploit imitation learning use simple positive and negative reinforcement, and place the burden of learning completely on ...
متن کاملGeneralizable Framework for Designing and Optimizing Dynamic, Robust, and Adaptive Bipedal Locomotion
Before robots can become a viable technology for assisting people with everyday tasks, they must be able to adapt to the ever-changing conditions found in real, human-centered environments, designed for a person’s ability to walk on two legs. This dictates the need for humanoid robots that can exhibit robust, bipedal locomotion. This thesis explores different methods for developing and optimizi...
متن کاملExploring the Lombard Paradox in a Bipedal Musculoskeletal Robot
Towards advanced bipedal locomotion musculoskeletal system design has received much attention in recent years. It has been recognized that designing and developing new actuators with the properties of the human muscle-tendon complex is only one of the many tasks that have to be ful lled in order to come close to the powerful human musculoskeletal system enabling the human to such versatile dyna...
متن کامل