Odometry During Object Transport: A Study with Swarm of Physical Robots

Abstract

AbstractObject transport by a single robot or by a swarm of robots can be considered a very challenging scenario for odometry since wheel slippage caused by pushing forces exerted on static objects and/or by relatively frequent collisions with other robots (for the cooperative transport case) tend to undermine the precision of the position and orientation estimates. This paper describes two sets of experiments aimed at evaluating the effectiveness of different sensory apparatuses in order to support odometry in autonomous robots engaged in object transport scenarios. In the first set of experiments, a single robot has to track its position while randomly moving in a flat arena with and without an object physically attached to its chassis. In the second set of experiments, a member of a swarm of physical robots is required to track its position while collaborating with the group-mates to the collective transport of a heavy object. In both sets, odometry is performed with either wheel encoders or with an optic-flow sensor. In the second set of experiments, both methods are evaluated with and without gyroscope corrections for angular displacements. The results indicate that odometry based on optic-flow sensors is more precise than the classic odometry based on wheel encoders. In particular, this research suggests that by using an appropriate sensory apparatus (i.e., an optic-flow sensor with gyroscope corrections), odometry can be achieved even in extreme odometry conditions such as those of cooperative object transport scenarios.KeywordsOdometryCooperative transportSwarm roboticsOptic-flow sensor