Garrett Jones, David C. Slaughter,
Biological & Ag. Engineering, UC Davis
1) real-time computer vision system
2) automate hydraulic control
3) low cost longitudinal sensor evaluation
4) user-friendly operator interface
1) 76,000 kilometers of highway stripe painting/year
2) 85 person-years of labor used per year
3) Driver and operator jobs tedious & fatiguing
1) Reduce impact on traffic flow
2) Improve efficiency of operation
3) Enhance safety
1) Generate a video data base of worn-out strips
2) Develop real-time computer vision algorithm
3) Build microcontroller to control painting
4) Research a low cost vision system
5) Retrofit existing hydraulic control system
6) Test reliability and precision of whole system
The current method of controlling paint sprayer nozzle involves placing a mirror on a 20 foot boom. The MB driver then positions the nozzle by steering the truck while focusing attention upon the boom mirror. The boom mirror represents a significant traffic hazard and limits MB truck speed.
Longitudinal control of the dashed line was evaluated using the existing encoded ground wheel and a radar sensor. The current encoded wheel generated 1.8 inches of standard error while the radar cause only 1 inch of standard error. The advantage of the radar is that is has no moving parts and thus reduces maintenance.
The challenges to the vision system include constantly varying lighting. Two cameras set for sun and shade conditions is one possible solution.
First prototype camera system mounted on the hydraulically actuated spray nozzle position arm.
last modified 12/17/2003