Obstacle Avoidance System for Autonomous Transportation Vehicle based on Image Processing

The labor shortage and the aging problem of farmers are serious problem for Japanese agriculture. In order to solve those problems, this research suggested that the smart mechanization system should play an important role for food production. The Autonomous Transportation Vehicles (ATVs) has a possibility to achieve this task. ATVs was applied with a CCD camera for vehicle control (i.e. for straight running, turning and emergency stop). We reported here how to recognize an obstacle for safe running by image processing. Hue, Saturation and Intensity were applied for obstacle recognition parameters and a binary image was produced by image processing. Finally, distance information (i.e. between vehicle and obstacle) was estimated using coordinate transformation between camera coordinate and vehicle coordinate. Experiments, indicated the obstacle recognition method could detect obstacles under the both sunny and cloudy condition. The error of distance estimation was within 0.3m and its processing speed was 5.9 frame/s.