Guidance for Specific Target Doors in Hallway using the Computer Vision for Autonomous Vehicles

Computer Vision would complement the autonomousrobotics goals of accuracy and precision by adding “eyes” to theperformance of the robot. We will be researching on computervision techniques and experimenting their benefits withautonomous robotics. We investigated both Arduino MegaTMController Board and Raspberry Pi for analyzing the flexibilityof chipset for computer vision using the 4-wire dual data busPixi Camera (CMU’s 5th version). Both C/C++ and Pythonlibraries were implemented to test the vision camera using I2Cand USB communication. Pixi camera scans at a rate of 50frames per second. We investigated the object recognition,object classification and object tracking capabilities usingvarious algorithms for detecting the specific target doors in ahallway. The autonomous robot is going to maneuver in the hallway by avoiding the obstacles and stop precisely at a door-frontof the assigned door number. This is possible by associating thedoor numbers with color codes.