Sensor-based robot motion generation in unknown, dynamic and troublesome scenarios

Abstract

A sensor-based motion control system was designed to autonomously drive vehicles free of collisions in unknown, troublesome and dynamic scenarios. The system was developed based on a hybrid architecture with three layers (modeling, planning and reaction). The interaction of the modules was based on a synchronous planner-reactor configuration where the planner computes tactical information to direct the reactivity. Our system differs from previous ones in terms of the choice of the techniques implemented in the modules and in the integration architecture. It can achieve robust and reliable navigation in difficult scenarios that are troublesome for many existing methods. Experiments carried out in these scenarios with a real vehicle confirm the effectiveness of this technique.

Decision-Theoretical Navigation of Service Robots Using POMDPs with Human-Robot Co-Occurrence Prediction

“…The wavefront-based global path planner calculates the optimal path and the reference points along the path based on mapped obstacles. The Nearness Diagram [12] based local reactive obstacle avoidance controller computes actual translational velocity  and rotational velocity  based on the reference points and real-time sensory data.…”

Modeling dynamic scenarios for local sensor-based motion planning

“…To overcome this limitation it has been suggested to combine these reactive techniques with some sort of planning (see [1] for a discussion on integration schemes and [30] for a similar discussion in the motion context). The more widespread way to combine reaction with planning are the systems of tactical planning [42,50,9,30,46,41,37].…”

“…The more widespread way to combine reaction with planning are the systems of tactical planning [42,50,9,30,46,41,37]. They perform a rough planning over a local model of the scenario, which is used to guide the obstacle avoidance.…”

Global optimal path planning of an autonomous vehicle for overtaking a moving obstacle

“…The cost functions depend on the problem and may be the time of travel or deviation from a reference path or any other dynamic or kinematic property of the vehicle. Several optimization methods for the path planning problem of robots have been applied in the past (Yahja et al, 2000;Arras et al, 2002;Spenko et al, 2004;Minguez and Montano, 2005;Rashid et al, 2013) and some have recently been used in the autonomous vehicle researches (Purwin and D'Andrea, 2006;Yoon et al, 2009;Zhe et al, 2009;Matveev et al, 2011).…”