An Online Planning Method Based on Kernel Sampling Strategy in Uncertain Environment

Abstract

The present invention proposes an online planning method based on the Kernel sampling strategy in an uncertain environment, which is used for planning when the robot performs a task. The main cause of operation; in the POMDP model, the robot can observe part of its own state, and the robot obtains the strategy with the greatest return by continuously interacting with the environment; in the online planning method, when dealing with the observable part, the robot’s The state is represented as a belief, recorded as belief, which belongs to a set of states, and the forward search is performed by constructing a belief tree with the POMDP algorithm, so as to obtain the optimal strategy under the current belief; each of the belief trees A node represents a belief, and a parent node and a child node are connected through a behavior-observation branch; the POMDP algorithm is an online POMDP planning algorithm Kernel-DESPOT; the algorithm performance of the present invention is better than that of DESPOT and POMCP, and has advantages in convergence speed and quality.

Description

technical field [0001] The invention relates to the technical field of robots, in particular to an online planning method in an uncertain environment based on a Kernel sampling strategy. Background technique [0002] With the rapid development of information technology, robots have gradually integrated into daily life. As an important field of robotics research, motion planning has received extensive attention. [0003] The sampling-based motion planning method converts the continuous model into a discrete model. Traditional sampling-based motion planning algorithms include RRT, PRM, and random potential field methods. The characteristic of the RRT algorithm is that it can quickly find a path in a relatively complex environment, but there are problems such as poor path quality, low efficiency in a narrow channel environment, and "blindness" in exploration. Both PRM and RRT are probabilistically complete and not optimal. PRM is essentially different from PRM in the arrangem...

Images