Trajectories of Multiple People in Crowds Using Laser Range Scanner

Chapter written by Ladji Adiaviakoye, Patrick Plainchault, Jean-Michel Auberlet. 

Traffic Safety, First Edition. Edited by George Yannis and Simon Cohen. ISTE Ltd 2016. Published by ISTE Ltd and John Wiley Sons, Inc. 

However, the high sensitivity to weather conditions, limited opening angle of the cameras and the inability to obtain accurate information on depth directly are major obstacles to the reliability and robustness of the system based on video camera. Indeed, many cameras are needed to cover a large area and also reduce occlusions due to objects or other people on the move. However, Bruckner et al. [BRU 11] explained that it is difficult to digitally merge data from multiple cameras, which requires precise calibration and high computational cost and complicated calibration. 

On the basis of that fact, we focus on an active sensor such as a laser scanner. This sensor is an instrument that has received increasing attention to the problems of monitoring. It measures the distances of nearby objects by emitting laser beams eye safety in controlled directions. To analyze this information on a distance a variety of technical processes are possible, (flight time) and depend on the magnitude of the distances measmed. Several authors, Prassler et al. [PRA 99], Schulz et al. [SCH 01], Mon et al. [MON 02], often use this type of sensor in many practical applications of robotics. 

This section first presents briefly the sensor system for tracking pedestrians. Emphasis is placed on the original operating data from our sensors. A description of the tracking algorithm starting with the choices made for the detection of pedestrians is given. Rao-Blackwellized Monte Carlo data association can track and identify objects in time with its various uncertainties. 

---

Trajectories of Multiple People in Crowds Using Laser Range Scanner... 

---