The implementation of an assistive robot with real-time image recognition functions

In general, a multifunctional image recognition system usually requires a lot of significant resources such as numerous memory components, rapid operation capability of CPU/GPU, image capture device or camera, and display devices/interfaces etc. Most of PC-based image recognition systems can handle complicated image processing. However, for some specific applications, the user will prefer to choose the easier and specific image recognition systems. It is not economic and efficient to do some easy works but using a lot of expensive resources. Therefore, developing a standalone microcontroller based image recognition system is a better choice for some specific applications. An assistive robot for the disabled with spinal cord injury was implemented in this research. The assistive robot system with the functions of real-time image recognition can deliver a small assistive pacifier (a limit switch inside) to the mouth of the disabled automatically. The disabled can control many things alone by using the assistive pacifier. The real-time image recognition subsystem of the automatic assistive robot is composed of two major parts: one is digital logic circuits built in FPGA module, the other is the microcontroller 8051. The digital logic circuits in FPGA module are designed to deal with real-time image processing. The microcontroller 8051 can be programmed to communicate with FPGA module and control the assistive robot to finish the task automatically. This essay also reveals a programmable color-tone table to make the image recognition system suitable for different environments with different light conditions by modifying color-tone tables. The successful image recognition rate of the experiments for verifying the performance of the assistive robot is over 93%.