Robot programming
In this project, I am working on the design and implementing an AI software that gives sorting robots the ability to autonomously generalize their knowledge and adapt to unexpected changes in environments.
Robotic arms are increasingly popular in industries due to their ability to boost production speed and minimize errors. They follow pre-written programs and are commonly used in structured work environments, like factory production lines, for tasks such as sorting objects. Recently, sorting robots integrated with vision systems have been introduced in markets like logistics, waste management, and recycling. However, challenges arise from unexpected environmental changes, making visual perception difficult for human programmers. To address this, more effective learning techniques are needed. Collecting extra data and experimenting with objects can help robots better handle real-world unpredicted situations, improving recognition and decision-making accuracy.
The software I have been developing is a control system for robot arms that allows them to observe and experiment with objects, hypothesize their type based on reasoning, and put these hypotheses to the ultimate test. The software has a hierarchical architecture that gives its reasoning unit the extracted visual information output from unsupervised object recognition algorithms. The robot programs are written in Lua programming language.
This is a demo of how my initial code for the arms controller works:
The software is under development. The ultimate goal is to connect AERA system, a general AI controller, to the robot, by which the arm’s controller learns to autonomously adapt to new environmental conditions, with minimum assistance from humans, and perform their tasks correctly. This adaptive intelligent control system for sorting robot arms can prepare them for different markets. The two major future markets for the software are warehousing and waste recycling industries, where robot arms can be employed to perform product/waste sorting tasks autonomously.