Search Results

A new concept for communication for platooning of leader-follower vehicles is presented. The leader vehicle is equipped by a omni directional camera and laser radar. The omni directional camera can capture a omni directional views and the view information is used to autonomously control the leader vehicle. The laser radar set to the rear part of the leader vehicle detects the relative distance and angle of the follower vehicle. By the information of the relative position and angle of the follower vehicle, control instructions are generated and sent via vehicle-to-vehicle communication between the leader vehicle and the follower vehicle.

The sensing and control scheme of autonomous vehicle, learned from the human driving is considered. To implement human driving manners, we imitate “characteristics of the human eye” and “human driving rules extracted from driver’s feeling”. Following two approaches are considered; 1. Reduction of image processing load of FFT picture compression technique by the manner of human eye sensitivity. 2. Simplified driving rules extracted from the human driving manner correspond to the energy minimum optimal control.

This paper proposes how to develop an autonomous vehicle for factory and/or storehouse use. Where the geometric arrangement changes frequently due to the turnover of the cargo. We apply the sensor fusion technique by use of the Omni directional image sensor and laser radar generate the map of such situations. We propose a new update algorithm called SMT (Shift Matching Transform) that obtains region segments with trusted semi-real-time geometric map. We demonstrate how to determined the absolute position of the vehicle by the position information of arbitrary wall. Experimental results on the corridor, Shows the semi-real-time map reconstruction with enough validity.

The project of the ITS (Intelligent Transport Systems) has been actively promoted researches and developments for the safe and efficient traffic. In the project, development of the lane detection algorithm is indispensable. Under outdoor environment, it is very difficult to detect a lane and to distinguish the various obstacles by vision systems, due to optical problems. Here we propose the Active Contour Model Method, which is known as the most robust method to detect the lane contour. The vehicle based on an electric wheelchair for handicapped persons was used to examine the validity of the method.

An automatically guided vehicle by vision that recognizes the two white guidelines has been developed. The vehicle can run on a mildly bumpy grass course like as golf course up to 3 km/hour, staying in between two white lines with span of 3m to 3.6m avoiding obstacles within lanes. In this paper, the construction of the vehicle, the sensing strategy, the control algorithm, as well as practical aspects of the implementation are described. Employment of 1D image sensor which detects two white lanes yields the remarkable feature of the proposed system. The advantage of 1-D image sensor is the simplicity as well as significant reduction of the task for image processing in comparison with the conventional 2D-camera vision. Through the actual experiments at International Ground Robotics Competition '97, which held at Oakland University in June, we demonstrated the superior performance compared with conventional 2D vision type vehicles.

This paper describes the accurate and detailed estimation of traffic conditions around our vehicle on road by the incomplete laser radar signals. The conventional laser radar system monitors only one vehicle that runs just in front of the our vehicle. Careful observation of the laser radar signals includes more information than that required. This research aims at developing a new traffic condition monitoring system that monitors 8 to 10 vehicles that run around our vehicle by fully using the signals from the radar.