Based On The Primary Obstacle Detection And Passive Vision Robot Local Path Planning

Ground intelligent mobile robot have been used more and more widely in the fileds of transportation,military,space exploration,industrial and so on,attracting scientific and technical personnel from all over the world to carry on in-depth study on it.Besides,obstacle detection and path plan are research focuses of the ground intelligent mobile robot,which have an important significance to promote robots to serve better in various fields.Lidar and camera are commonly used sensors.Lidar is a generalized active visual sensor with a small amount of data and high precision.However,it has limited detection range.Camera is a typical passive visual sensor which can extrace color and shape of obstacles,but it can not obtain the distance information of obstacles directly.This paper researches on obstacle detecting by clustering lidar data and image segmentation,then fuses laser data and camera data.First,project lidar data to the image as seeds,carry on region growing and get credible obstacle region.Second,compare with suspicious obstacle region obtained by image segmentation,add the undetected suspicious obstacles to the re sult, thereby obtain all-sided and reliable environment information for further processingDue to the shortcomings of principle,Iocal path planning methods such as artificial potential filed method and reactive method exsit local minimum or other problems.The research trend is to combine global and local path planning methods,or combine several local path planning methods,in oreder to overcome the shortcomings of single one.This paper analyzes the artificial potential field method and reactive mehod,then proposes a local path planning strategy based on circular trajectory.First,we use GPS and two-point method for global path planning.Second,we use the scroll window method to update the current window and search for local optimized sub-goals.Finally,we use local path planning strategy based on circular trajectory to reach the sub-goals.It’s more corresponding to the kinematics of ground intelligent mobile robot,which can get a smooth trajectory as well.The evaluation shows the superiority of path planning based on circular trajectory compared to mehod based on straight-line trajectory.However,the single circlular trajectory commonly used has the drawback that the robot change direction when it reaches the goal and can not observe effectively. In this paper,we propose a path planning strategy based on dual cicular trajectory according to circular trajectory.The experiments show that this strategy overcomes the shortcoming of single circular trajectory while remains the superiority of the circular trajectory.