Research On The Falling Detection Algorithm Of Pedestrian Targets In Mobile Perception Robots

With the rapid increase in the number of people living in the elderly population,reducing and dealing with the problem of falls in the elderly has become the focus of research for decades.It is impossible to completely eliminate falls in daily life and activities.Detecting a fall in time can protect the elderly from injury as much as possible.Therefore,this subject develops a robot that builds an environment map in an indoor scene and detects the fall of the elderly.The goal is achieved when the robot can map and navigate,avoid obstacles in space,detect the fall of an elderly person,and communicate when it detects a fall.This paper mainly studies the algorithmic problem of the fall detection of pedestrian targets in mobile sensing robots.The hardware platform,robot operating system,and robot kinematics modeling are detailed.It introduces in detail the simultaneous positioning and map construction algorithms for map building and navigation,adaptive Monte Carlo positioning,and compares the breadth-first search of global path planning algorithms,Dijkstra algorithm and A* algorithm,and introduces local paths Planning algorithm DWA.Finally,the ROS robot operating system is used on the Turtlebot robot,which combines simultaneous positioning and map construction technology,Monte Carlo positioning,A* path planning,dynamic window method and indoor map navigation.The network model,loss function and development of YOLO of the YOLO model are introduced.The robot perception algorithm was tested,and the YOLOv4 network was trained using the stance and fall data sets.The results were tested with three indicators: average precision,precision,and recall.In this paper,the precision is defined as The ability to correctly identify fall events,and the recall rate is defined as the ability to correctly exclude non-fall events.