The role of multi-agent systems in our lives
By Billy Lau Pik Lik and Dr. Ashutosh Kumar Singh
The ongoing exponential advancement of technology in today’s world has brought about greater automation and the reduction of human input in many spheres of life.
In the not too distant future, it is possible that we will depend on intelligent robots to carry out mundane tasks such as performing household chores.
According to The New York Times, years of research into robotics has paved the way for robots to step into our daily lives at any time. However, designing an intelligent robot that can cooperate and communicate with other robots remains a complicated endeavour.
Whilst computer scientists across the world have dedicated themselves to improving sections of distributed artificial intelligence for decades, one of their main challenges is getting intelligent robots to form groups or ‘coalitions’ in which they cooperate when required and dissolve when they are no longer needed. An analogy of such cooperation is companies grouping together to achieve certain goals or maximising profits.
Getting robots to manoeuvre in group formation is another challenge that involves a combination of hardware and software functioning in a mechanical way. However, the Massachusetts Institute of Technology has achieved a milestone in this area by getting quadcopters to fly in formation.
The main challenge in the development of current robotic and multi agents systems is the exorbitant cost involved. Development of an intelligent robotic product requires in-depth expertise in the robotics field as well as a significant amount of investment. Even a ‘cheap’ exotic robotic pet like Dinosaur Pleo costs USD469 (RM1,500).
The efficiency of hardware machines is another factor slowing down the advancement of such robots. The current agile robots available for industries are bulky and expensive, and therefore, not convenient for daily use.
Scientists are however making headway in overcoming the existing limitations and their discoveries will constitute great leaps in human technology. For instance, a logistics company can implement multi-agent systems into its delivery operation by installing GPS and various sensors to assist unmanned vehicles to travel on the road without any human effort. Each vehicle will be able to travel by land, air or sea to deliver packages to chosen destinations.
For each transaction arranged by the company, the software will assign vehicles to form a group to deliver the packages. Each vehicle will have its own package delivery plan with optimal routes planned.
Another example of application is in future warfare where the use of automated robots or unmanned vehicles could help reduce human casualties. The cooperation between robots in this aspect can affect the flow of battles. Therefore, it is important to create highly intelligent robots that can adapt to any situation rationally.
While it has been theorised that groups of robots could very well outperform humans in modern warfare without sacrificing the lives of soldiers, it has yet to be thoroughly researched and mathematically proven. Here again, there are significant challenges slowing progress of the development of the required software, including a large domain problem and difficulty in developing a universal solution.
Also, communication between the robots would be intense as they will need to synchronise their locations, transfer packages and update systems. A specific protocol for agents to identify their partners to form coalitions has yet to be designed and implemented.
Other factors such as high-speed channels that can sustain the intensity of the programme are also to be considered. Robots for military and medical purposes require more precise hardware and efficient software to perform real-time decisions to determine optimal strategy for eliminating enemies and saving lives.
In a nutshell, the foundation of artificial intelligence has contributed greatly to the future of modern technologies. For better solutions to technological problems, a strong foundation in artificial intelligence is needed and should be one of the main research fields in Malaysia.
It is essential to provide software solutions for hardware as intelligent software can reduce a lot of human effort. Furthermore, cheap hardware can also be used in developing advance robots provided the software is mature enough to drive the hardware.
Billy Lau Pik Lik is a Higher Degree Research (HDR) student in the Department of Electrical and Computer Engineering, School of Engineering and Science, at Curtin Sarawak. His research interests include multi-agent system interactions, organisation structures and based simulation as well as mobile security.
Dr. Ashutosh Kumar Singh is an Associate Professor and Head of the Department of Electrical and Computer Engineering. His research interests include verification, synthesis, design and testing of digital circuits and he has published over 70 research papers on the subjects in various conference and research journals. He co-authored two books, ‘Digital Systems Fundamentals’ and ‘Computer System Organisation & Architecture’, and has delivered talks on computer engineering in several countries including Australia, the United Kingdom and the United States. Dr. Ashutosh can be contacted by e-mail to email@example.com.