Decentralized Structural-RNN for Robot Crowd Navigation with Deep Reinforcement Learning.
Shuijing LiuPeixin ChangWeihang LiangNeeloy ChakrabortyKatherine Driggs CampbellPublished in: ICRA (2021)
Keyphrases
- reinforcement learning
- robot navigation
- obstacle avoidance
- mobile robot
- real robot
- recurrent neural networks
- robot control
- navigation tasks
- indoor environments
- service robots
- rough terrain
- multi robot systems
- multi agent
- nearest neighbor
- autonomous navigation
- unknown environments
- potential field
- perceptual aliasing
- path planning
- autonomous learning
- autonomous robots
- multi robot
- structural information
- autonomous vehicles
- function approximation
- visual navigation
- human robot interaction
- continuous state
- cooperative
- vision system
- mobile robotics
- simulated robot
- collision free
- robotic control
- reinforcement learning algorithms
- humanoid robot
- distributed systems
- robotic systems
- outdoor environments
- crowd sourced
- machine learning
- state space
- motion planning
- peer to peer
- dynamic environments
- neural network
- navigation systems
- decentralized control
- optimal policy
- markov decision processes
- modular robots
- back propagation
- optimal control
- multiple robots
- dead reckoning
- multiagent reinforcement learning
- temporal difference
- dynamic programming
- learning process
- formation control
- learning algorithm
- topological map