Fatal bus and pedestrian collisions have increased dramatically recently due to the distractions from various portable electronic devices. The goal of this project is to investigate various sensors for collision detection and propose efficient warning approaches for bus operators and pedestrians. Current solutions use repeating audio warning to alert pedestrians and bus operators. They measure steering column for turning detection, which requires extra devices installed on the vehicle steering system and broadcast message alerts with no consideration of the actual situations. GPS signal and onboard IMU (Gyroscope and Accelerometer) will be integrated to analyze the bus motion during turning. These sensors together with the computation component require no change on the basic vehicle structure and present actual motion information for crisis prediction. Various sensor technologies, such as laser detector, sound detector and regular cameras, will be evaluated to provide better detection and distance estimation of pedestrians. Once we have enough knowledge of the situation, proper mapping and planning algorithms will be studied to estimate collisions and to trigger a warning message. Apart from audio, warning messages will also be delivered to pedestrians through other channels, such as laser scanning and air curtains. For bus drivers, more specific guidance will be addressed to avoid collision based on estimation.
-
Sep. 2012 to July 2013
-
Impacts/Benefits of Implementation (actual, not anticipated) The ARM based embedded systems fits well in this task because of its low cost and low power requirement. IMU based the algorithm can measure the velocity and orientation standalone and generate bus trajectory without any modification on the bus system. GPS and IMU based detection algorithm senses bus turning beforehand and provide evidence to trigger warning. LIDAR based pedestrian detection system on the other hand, is able to find any obstacle front in certain range and generate a real time occupancy map of current environment. LIDAR detection also has the capability to work in night and extreme weathers.
Name | Affiliation | Role | Position | |
---|---|---|---|---|
ddlee@seas.upenn.edu | Lee, Daniel | University of Pennsylvania | PI | Faculty - Research/Systems |
Type | Name | Uploaded |
---|---|---|
Progress Report | 108_-_TSET_Report_.pdf | July 10, 2018, 8:34 a.m. |
Final Report | 108_-_TSET_Final_Report.pdf | July 11, 2018, 3:49 a.m. |
No match sources!
No partners!