Local Navigation Among Movable Obstacles with Deep Reinforcement Learning

Autonomous robots would benefit a lot by gaining the ability to manipulate their environment to solve path planning tasks, known as the Navigation Among Movable Obstacle (NAMO) problem. In this paper, we present a deep reinforcement learning approach for solving NAMO locally, near narrow passages. We train parallel agents in physics simulation using an Advantage Actor-Critic based algorithm with a multi-modal neural network. We present an online policy that is able to push obstacles in a non-axial-aligned fashion, react to unexpected obstacle dynamics in real-time, and solve the local NAMO problem. Experimental validation in simulation shows that the presented approach generalises to unseen NAMO problems in unknown environments. We further demonstrate the implementation of the policy on a real quadrupedal robot, showing that the policy can deal with real-world sensor noises and uncertainties in unseen NAMO tasks.

A Fully-autonomous Framework of Unmanned Surface Vehicles in Maritime Environments using Gaussian Process Motion Planning

Unmanned surface vehicles (USVs) are of increasing importance to a growing number of sectors in the maritime industry, including offshore exploration, marine transportation and defence operations. A major factor in the growth in use and deployment of USVs is the increased operational flexibility that is offered through use of autonomous navigation systems that generate optimised trajectories. Unlike path planning in terrestrial environments, planning in the maritime environment is more demanding as there is need to assure mitigating action is taken against the significant, random and often unpredictable environmental influences from winds and ocean currents. With the focus of these necessary requirements as the main basis of motivation, this paper proposes a novel motion planner, denoted as GPMP2*, extending the application scope of the fundamental GP-based motion planner, GPMP2, into complex maritime environments. An interpolation strategy based on Monte-Carlo stochasticity has been innovatively added to GPMP2* to produce a new algorithm named GPMP2* with Monte-Carlo stochasticity (MC-GPMP2*), which can increase the diversity of the paths generated. In parallel with algorithm design, a ROS based fully-autonomous framework for an advanced unmanned surface vehicle, the WAM-V 20 USV, has been proposed. The practicability of the proposed motion planner as well as the fully-autonomous framework have been functionally validated in a simulated inspection missions for an offshore wind farm in ROS.