A Low Cost Multi Sensorial Data Fusion for High Speed Obstacle Avoidance Using 3-D Point Clouds and Image Processing in Self Balancing Robots

摘要

This paper investigates the challenges related to a very fundamental problem of efficient obstacle avoidance on roads in the modern world, with respect to the self balancing robot. The amalgamation of navigation in self balancing robot pose newer challenges in circumvention of traffic congestion. Comprehensive research has been done on the obstacle avoidance of the four wheeled vehicles, but we have other problems to address in the case of self balancing robots on two wheels. Speed is becoming a predominant factor in the present day automotive industry. In this paper we address this very issue and propose a model driven by the outputs from multi sensorial data generated using various sensors and image processing techniques. The distinguishing feature of our technique is how we tackle the high speed obstacles. The paper channelizes its focus on avoiding high speed obstacles by tracking the object in real time using image processing techniques and then creating a 3-D point cloud of the object and its static surroundings through a matrix of arrays, using the Light Detection And Ranging (LiDAR) module.