Guillaume Bellegarda
Guillaume Bellegarda
  • Видео 9
  • Просмотров 3 794
Learning Human-Robot Handshaking Preferences for Quadruped Robots
Quadruped robots are showing impressive abilities to navigate the real world. If they are to become more integrated into society, social trust in interactions with humans will become increasingly important. Additionally, robots will need to be adaptable to different humans based on individual preferences. In this work, we study the social interaction task of learning optimal handshakes for quadruped robots based on user preferences. While maintaining balance on three legs, we parameterize handshakes with a Central Pattern Generator consisting of an amplitude, frequency, stiffness, and duration. Through 10 binary choices between handshakes, we learn a belief model to fit individual prefere...
Просмотров: 128

Видео

Visual CPG-RL: Learning Central Pattern Generators for Visually-Guided Quadruped Locomotion3:00Visual CPG-RL: Learning Central Pattern Generators for Visually-Guided Quadruped Locomotion
Visual CPG-RL: Learning Central Pattern Generators for Visually-Guided Quadruped Locomotion
Просмотров 1715 месяцев назад
We present a framework for learning visually-guided quadruped locomotion by integrating exteroceptive sensing and central pattern generators (CPGs), i.e. systems of coupled oscillators, into the deep reinforcement learning (DRL) framework. Through both exteroceptive and proprioceptive sensing, the agent learns to coordinate rhythmic behavior among different oscillators to track velocity command...
Quadruped-Frog: Rapid Online Optimization of Continuous Quadruped Jumping3:02Quadruped-Frog: Rapid Online Optimization of Continuous Quadruped Jumping
Quadruped-Frog: Rapid Online Optimization of Continuous Quadruped Jumping
Просмотров 5035 месяцев назад
Legged robots are becoming increasingly agile in exhibiting dynamic behaviors such as running and jumping. Usually, such behaviors are either optimized and engineered offline (i.e. the behavior is designed for before it is needed), either through model-based trajectory optimization, or through deep learning-based methods involving millions of timesteps of simulation interactions. Notably, such ...
CPG-RL: Learning Central Pattern Generators for Quadruped Locomotion6:06CPG-RL: Learning Central Pattern Generators for Quadruped Locomotion
CPG-RL: Learning Central Pattern Generators for Quadruped Locomotion
Просмотров 1,6 тыс.Год назад
In this letter, we present a method for integrating central pattern generators (CPGs), i.e. systems of coupled oscillators, into the deep reinforcement learning (DRL) framework to produce robust and omnidirectional quadruped locomotion. The agent learns to directly modulate the intrinsic oscillator setpoints (amplitude and frequency) and coordinate rhythmic behavior among different oscillators....
Dynamic Vehicle Drifting with Nonlinear MPC and a Fused Kinematic-Dynamic Bicycle Model1:34Dynamic Vehicle Drifting with Nonlinear MPC and a Fused Kinematic-Dynamic Bicycle Model
Dynamic Vehicle Drifting with Nonlinear MPC and a Fused Kinematic-Dynamic Bicycle Model
Просмотров 4092 года назад
Dynamic Vehicle Drifting with Nonlinear MPC and a Fused Kinematic-Dynamic Bicycle Model Abstract: In this letter we present a versatile trajectory optimization framework that leverages a fused kinematic-dynamic bicycle model for highly dynamic vehicle drifting maneuvers. Our framework can be used online to generate drifting maneuvers, offline to plan drift parking, and additionally enables onli...
Versatile Trajectory Optimization for Dynamic Vehicle Maneuvers1:13Versatile Trajectory Optimization for Dynamic Vehicle Maneuvers
Versatile Trajectory Optimization for Dynamic Vehicle Maneuvers
Просмотров 974 года назад
Versatile Trajectory Optimization for Dynamic Vehicle Maneuvers
Combining Benefits from Trajectory Optimization and Deep Reinforcement Learning3:00Combining Benefits from Trajectory Optimization and Deep Reinforcement Learning
Combining Benefits from Trajectory Optimization and Deep Reinforcement Learning
Просмотров 4474 года назад
This video shows Bullet racecar simulations of executing: -a policy learned with vanilla Proximal Policy Optimization (PPO) -a conservative trajectory optimization run as MPC -our combined method, CoTO-PPO, which enjoys benefits from both reinforcement learning and model-based trajectory optimization Details can be found in the paper "Combining Benefits from Trajectory Optimization and Deep Rei...
Trajectory Optimization for a Wheel-Legged System for Dynamic Maneuvers that Allow for Wheel Slip3:32Trajectory Optimization for a Wheel-Legged System for Dynamic Maneuvers that Allow for Wheel Slip
Trajectory Optimization for a Wheel-Legged System for Dynamic Maneuvers that Allow for Wheel Slip
Просмотров 2154 года назад
This video shows visualizations in MATLAB and simulations in MuJoCo of trajectories for JPL's Robosimian found by our optimization framework for various tasks (different objective functions). These include energy-efficient forward locomotion, a dynamic parking skidding maneuver, and hybrid wheel-legged skating. Details can be found in a preliminary version of the paper below, with the final ver...
Training in Task Space to Speed Up and Guide Reinforcement Learning3:00Training in Task Space to Speed Up and Guide Reinforcement Learning
Training in Task Space to Speed Up and Guide Reinforcement Learning
Просмотров 2775 лет назад
This video shows simulations of executing a policy learned with Proximal Policy Optimization (PPO) for various tasks on JPL's Robosimian. More details can be found at arxiv.org/abs/1903.02219

Комментарии