An autonomous mobile robot system with adaptive navigation strategy and vision-based motion planning
R. Lal Tummala, Cheng‐Chih Lin · 1994
Autonomous mobile robots have drawn much attention in both academic research and industrial applications because of their intelligent behavior and versatility. These robots utilize various types of sensors to perceive their environments, and use them to perform motion planning. This dissertation addresses the problems of sensor-based navigation in typical manufacturing environments that are structured, partially known, and dynamic. The first half of this work focuses on an adaptive navigation strategy, where the speed and accuracy constraints during a navigation process are treated as functions of the changes in the robot's environment. The second half introduces a new vision-based mobile robot motion planning system which can be integrated with the dead-reckoning method to achieve accurate goal acquisition. At the end, a hybrid navigation approach is developed to exploit the flexibility of adaptive navigation strategy and the advantages of vision-based motion control. This work can be summarized as follows: (1) Proposed a Weighted Obstacle Density Function (WODF) for the quantitative measurement of the clutterness of the robot's workspace. (2) Developed a quantitative measurement of the quality of general map construction algorithms, namely the Match Indices. (3) Developed the Adaptive Navigation Strategy (ANS) for a mobile robot. This approach is based on adjusting sensor configuration and motion speed of the robot according to the clutterness of the environment. (4) Developed a vision-based self-calibration system using circular disk landmarks. Designed a visual-servo motion control system for mobile robot docking operation. (5) Proposed a hybrid navigation system to integrate the ANS and the vision-based motion planning approach.