An intelligent computer vision integrated regression based navigation approach for humanoids in a cluttered environment

With a clear edge over the mobile robot counterparts, humanoids have become the centre of attraction for people dealing with robotics research. The enhanced use of the humanoids in industrial automation, manufacturing and other related areas has forced researchers to focus on their navigational aspects. In the current work, a computer vision integrated regression based navigational approach has been designed and implemented on a humanoid. Initially, a regression control architecture has been formulated for path planning and obstacle avoidance of a humanoid considering sensor information regarding obstacle distances as the inputs and the necessary heading angle as the output for the controller. Then, the limitations available in the regression based approach have been found out. To avoid the limitations, a computer vision based technique has been integrated with the original regression based approach. It has been observed that by the use of computer vision based technique, the robot is able to clearly distinguish between different types of obstacles, arena and target and reach the target position safely. Multiple simulations and real-time experiments have been conducted to verify the effectiveness of the proposed controller. The results obtained from both the simulation and experimental platforms have been compared against each other in terms of navigational parameters, and a good agreement between them is observed. The developed approach has also been assessed against another existing navigational technique, and a significant performance improvement has been observed. Finally, concluding remarks have been given regarding the use of both the techniques in humanoid navigation in complex environments.