Cooperative Sensor Sharing for Shared Situational Awareness, 10-R9784Printer Friendly Version
Inclusive Dates: 02/11/08 12/31/08
Background - SwRI and the Institut National de Recherche en Informatique et en Automatique (INRIA), a French national institute that conducts fundamental and applied research in information and communication science and technology, signed a collaborative agreement September 5, 2007, to conduct joint research and exchange intellectual property and staff to foster rapid technology and system advancements in vehicle autonomy. As part of this collaboration, an SwRI staff member spent three months working in the laboratories at INRIA, located near Versailles, France. The joint research undertaken on this project addresses the mutual benefits of sharing context-specific sensor data between vehicles, so called vehicle-to-vehicle (V2V) communication, specifically for pedestrian protection. The concept of shared situational awareness (SSA) was developed as a premise for this type of cooperative sensor sharing among vehicles.
Approach - The scenario for which the SSA system was developed is a pedestrian crossing in a crosswalk against traffic in front of a parked bus or van. A vehicle traveling in a direction where the pedestrian is visible detects the pedestrian and alerts other vehicles in the area via a dynamic short range communications (DSRC) radio. C++ classes were created to support collecting, formatting, reading, and analyzing the shared sensor data between vehicles. The relevant data that is transmitted is the sender's GPS position, the pedestrian's GPS position in latitude and longitude, the pedestrian's speed and heading, and the location of the crosswalk. The receiving vehicle first determines the relevance of the data by evaluating the relative position of the sending vehicle and pedestrian. The information is disregarded if the information concerns a crosswalk or pedestrian that is "far" away or behind the receiving vehicle. If the information is determined to be valid, the receiving vehicle then determines the distance to the pedestrian, and if the distance is within a reaction-distance threshold, the vehicle will slow and stop at the location of the crosswalk. This functionality was developed for both the INRIA Cycab and the SwRI Mobile Autonomous Robotics Technology Initiative (MARTI®) autonomous vehicle development platforms.
Accomplishments - A demonstration of two Cycabs cooperating to avoid a pedestrian collision was performed at INRIA in June 2008. One Cycab was equipped with a laser incident detection and ranging (LIDAR) sensor, differential GPS and WiFi wireless communication. It was positioned at the edge of a pedestrian crosswalk, described in the software as a GPS latitude/longitude location. A second Cycab was equipped only with GPS and wireless communication, and it drove autonomously in an adjacent lane, passing the parked Cycab before entering the crosswalk. A pedestrian would cross back and forth in the crosswalk, and as the "blind" Cycab approached the crosswalk it would slow and then stop. When the pedestrian had left the crosswalk, the Cycab would resume its travel.
The software algorithms were then ported to the MARTI vehicle, and customized for the vehicle's LIDAR and GPS sensors as well as the DSRC radio. The demonstration scenario was also adapted for the Intelligent Transportation Systems World Congress, which was held in November 2008 in New York City. The algorithms were also improved so that a generic crosswalk could be loaded from GPS data and the pedestrian's position relative to the direction of traffic could be evaluated.