Double-vfh: Reliable Obstacle Avoidance for Large, Non-point, Omni-directional Mobile Robots

This paper presents a method for reliable real-time obstacle avoidance for a tele-autonomous omnidirectional mobile robot, called “OmniMate.” The OmniMate is a large, rectangular-footprint platform with a flat loading deck measuring 184 ×92 cm (72×36”). Implemented with a shared control architecture, the OmniMate can operate in a so-called tele-autonomous mode that allows the operator to prescribe a general direction of motion even without visual contact. If the OmniMate encounters an obstacle, it autonomously avoids collision with that obstacle while trying to match the prescribed direction as closely as possible. Tele-autonomous operation is based on the newly developed Double-VFH (DVFH) method, which is a modified version of our earlier Vector Field Histogram (VFH) method. The DVFH method accounts for the rectangular footprint of the OmniMate by calculating simultaneously two vector field histograms centered at two points along the longitudinal axis of the OmniMate, based on the environment information dynamically derived from a ring of 32 ultrasonic sensors. Experimental results of the OmniMate traversing a densely cluttered obstacle courses at an average speed of 0.2 0.3 m/sec and a maximum speed of 0.5 m/sec are also presented.