Planning under uncertainty in the continuous domain: A generalized belief space approach

Abstract

Abstract-This work investigates the problem of planning under uncertainty, with application to mobile robotics. We propose a probabilistic framework in which the robot bases its decisions on the generalized belief, which is a probabilistic description of its own state and of external variables of interest. The approach naturally leads to a dual-layer architecture: an inner estimation layer, which performs inference to predict the outcome of possible decisions, and an outer decisional layer which is in charge of deciding the best action to undertake. The approach does not discretize the state or control space, and allows planning in continuous domain. Moreover, it allows to relax the assumption of maximum likelihood observations: predicted measurements are treated as random variables and are not considered as given. Experimental results show that our planning approach produces smooth trajectories while maintaining uncertainty within reasonable bounds.

SLAM and Vision-based Humanoid Navigation

“…The presentation decoupled localization and planning as it is currently the most common approach. However it is interesting to note that there is a growing number of works coupling estimation and model predictive control using a unified approach [24,15]. The main difficulty is probably to perform efficient implementation able to cope with the large number of variables of the generalized state.…”

Comparison of two strategies of path planning for underwater robot navigation under uncertainty

“…However, it is very difficult to quantify the weight coefficient without prior knowledge. If the weight coefficients are fixed, the robot may stay at some region that locally minimize (8). Considering the main objective is that the underwater robot arrives at the destination as soon as possible, we devise a self-adaptive strategy to address this problem.…”

“…The solution * of the optimization problem (8) is the optimal action between the time step and the time step + 1.…”

“…Roijers et al [7] propose a motion planner based on multi-objective Markov Decision Process, which considers the motion uncertainty but ignores the observation uncertainty. Recently, Vadim et al [8] propose a planning approach including the estimation of the position of landmark. These approaches are based on the Discrete Stochastic Process, which regards robot's motion as sequential actions that are not deterministic.…”

Planning in the continuous domain: A generalized belief space approach for autonomous navigation in unknown environments

“…The present paper is an extension of the work presented by Indelman et al (2013) and Indelman et al (2014). As a further contribution, in this manuscript we present an extensive experimental evaluation (Section 7), with the aim of testing the proposed approach in a large variety of challenging scenarios.…”