Experiments in Real-Time Spatial Data Acquisition for Obstacle Detection

Abstract

Rapid spatial data collection is desirable for many applications in engineering and construction. Recent technology developments such as Flash LADAR devices (Laser Detection and Ranging) allow rapid acquisition of range and intensity scene data. Methods that process the ouput into 3D side models and display or provide the location of objects on a site signif icantly improve work processes. A particularly useful application is modeling the presence of obstacles in structured or unstructured construction-type scenes including people, material, and equipment. The development of obstacle avoidance systems improves safety during the operation of equipment and efficiency for pick ing and plac ing objects. This paper describes laboratory experiments and field trials under realistic conditions for the development of obstacle detection algorithms for construction machinery using a high frame -rate Flash LADAR sensor. The focus is on the acquisition of experimental data to determine the benefits and limitations of this approach within the construction industry. Tests were conducted to analyze and discuss the sensor performa nce and evaluate the trade -off between sce ne detail and processing speed. Future research goals are in the development of algorithms allowing accurate scene modeling by acquir ing data from a moving object and detecting both stationary and moving obstacles in the motion path.