CAR-DESPOT: Causally-Informed Online POMDP Planning for Robots in Confounded Environments.

Ricardo Cannizzaro Lars Kunze

Published in: IROS (2023)

Keyphrases

robot teams
planning problems
robotic systems
multi robot
mobile robot
belief space
online learning
autonomous robots
partially observable markov decision processes
multiple robots
dynamic environments

partially observable markov decision process
partially observable
uncertain environments
motion planning
decision theoretic
unstructured environments
belief state
heuristic search
state space
real time
reinforcement learning

point based value iteration
service robots
collision free
stochastic domains
planning under uncertainty
autonomous systems
robot control
path planning
markov decision processes
predictive state representations

partially observable stochastic domains
search and rescue
partial observability
markov decision problems
physical objects
reward function
humanoid robot
cooperative
multi agent
real world