Study On The Methods Of Obstacle Detection And Road Tracking Based On Mobile Scanning Downward-Looking 2D Lidar

Recently,outdoor autonomous navigation robots have been widely used in different application areas,such as small unmanned ground vehicles,delivery robots.Obstacles and road feature detection is an important task of the outdoor autonomous navigation robots in robot applications.Lidar is widely used to outdoor navigation of mobile robots because of its high speed and accuracy.Lidar is divided into single-line(2D)Lidar and multi-line(3D)Lidar.Single-line Lidar is more widely used in small outdoor mobile robots,because of its economic advantages.In this thesis,a downward-looking 2D Lidar is used to detect obstacles and road boundaries in front of the robot while it is moving forward.The robot travels along the right side of the road according to the detected boundary.The main contents are as follows:Firstly,extracting line segments from each scanning data and presenting an adaptive obstacle detection method.The method estimates the height and the vector of the scanned road surface at each moment,and the segments are divided into either road ground or obstacles base on the average height of each line segment and the deviation between the line segment and the road vector is estimated from the previous measurement.The estimated road height and road vector can vary with the changing road conditions and reflect the road condition at different moment,which improves the adaptivity of our method.Secondly,in order to extract the safe driving area and the driving direction of the robot,the road boundary information is extracted.Because it is difficult to accurately detect road boundary information with one frame of data.In this paper,the method of multi-frame data fusion is adopted.The candidate points of road boundary are extracted by first-order differential filtering and weighting for each frame of Lidar scanned data,and we use road boundary information for boundary fitting,then the boundary is tracked by Kalman filter.Finally,the current obstacles and road boundary information are mapped to the local grid map of the robot,and a method of target guidance points is proposed.Combining with A*algorithm,path planning is carried out in this local grid map to enable the robot to travel along the right road boundary.The validity of the above algorithm is verified through a series of experiments,and the robot can effectively avoid obstacles along the boundary.