Visual Based Distance And Speed Estimation For Collision Avoidance

Wan Norzuraidi, Wan Nurfadzlyn (2019) Visual Based Distance And Speed Estimation For Collision Avoidance. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (24 Pages) Visual Based Distance And Speed Estimation For Collision Avoidance.pdf - Submitted Version Download (324kB)

Abstract

Estimation of distance and speed of an object to a stereo camera system have feasible advantages among different applications in various field. Autonomous vehicles and robotics often require distance and speed estimation for collision avoidance. This paper introduces a system to estimate the distance and speed of an object from two 5MP USB cameras. Calibration of stereo cameras are done beforehand using MATLAB Stereo Calibrator Application. The results of the calibration is analysed and used to measure the speed and distance of an object. The accuracy of this suggested approach is calculated and compared using appropriate tool and formula. Triangulation and Euclidean formula is used to measure the distance between object to cameras. Meanwhile, speed of the object is calculated using linear velocity formula at a specified time of 2 seconds. From the result obtained, the average percentage error for 0.05m, 0.06m and 0.07m distance between two cameras are 3.05%, 6.97% and 1.93 % respectively. Accordingly, the average percentage error for speed estimation data is 5.77%.

Actions (login required)

View Item