Evaluation of Human Intervention-Based Hybrid Approach for Position and Depth Estimation With Error Correction

摘要

Position and depth (PD) estimation is one of the key characteristics of autonomous robots. Robots are often challenged to visualize an alien environment remotely, alongside the control mechanism, and estimate the PD of objects. For the robot to reach out to an object, it needs to know the object’s position in a 3-D space. The design of the robot’s vision system is crucial. In this research work, we propose a human intervention-based hybrid approach for estimation of PD of an object. Human intervention in the form of a mouse click on the laser spot of the object image/in the video created by a camera-laser setup is used to estimate the PD of an object. An error correction model is developed and evaluated for better performance of the proposed method. A comparison of the proposed method with that of the image processing method revealed that a hybrid approach is almost 50% better in accuracy. The test results indicate that this method could be very helpful in finding the depth of the object with better accuracy in a teleoperated semiautonomous robot.