Creating cooperative robots with intelligence capabilities

Citation

Liu, H. (2008), "Creating cooperative robots with intelligence capabilities", Industrial Robot, Vol. 35 No. 1. https://doi.org/10.1108/ir.2008.04935aaa.002

Creating cooperative robots with intelligence capabilities

Viewpoint

Creating cooperative robots with intelligence capabilities

Today's robot-assisted industrial systems face a rapidly changing market due to the global economy and advanced information technologies. There are many new challenges in such systems design, development, implementation and improvement. Fast deployment of industrial requirements and tasks into small scale or streamlined robotic systems is a core competence for sustainable manufacturing and the other robot-oriented industries such as health care.

Robotics research has traditionally emphasized low-level sensing and control tasks including robot design, trajectory planning, sensory processing and control strategies, and it has been applied in a variety of application domains with industrial recognition over the past four decades. For instance, assembly robots in the automotive industry have been undertaking a wide range of tasks from simple part assembly to complex stamping and assembly-by-welding. However, almost all robot-related industrial applications such as streamlined assembly have been achieved through hard coding, offline specific testing and online numeric control. It significantly lacks flexibility for fast deployment, efficient production and quality service. Not only does this limitation significantly affect the target of low-cost operational efficiency production management, but also it is the route for repetitive and boring labor tasks with serious affects on health and safety issues. The author has witnessed the reality in streamlined assembly in the automotive industry. These applications urgently require robots and automation with intelligent capabilities. This means that an industrial robot should have the capabilities of integrating fundamental reasoning, perception and action with conventional industrial tasks and complex operations.

The use of multiple cooperative robots offers significant advantages over single robots in such industrial tasks as object grasping, material handling, assembly tasks and heavy load manipulation, etc. Owing to recent innovations in scientific methods and technologies; for example, with sensor technology and off-the-shelf algorithms in computational intelligence, it is now feasible and practical to enable a group of robots with intelligent capabilities in civilian and military applications, ranging from healthcare delivery, automotive assembly to space exploration. Such cooperative robotic systems are expected to have a substantial global impact on the way in which service or production systems, especially in the healthcare industry, are designed and operated.

The term of intelligence capability in contrast to computer numeric control of conventional robotic systems. Recent innovation in computational intelligence including fuzzy learning, neural networks, evolutionary computation and classical artificial intelligence provides sufficient theoretical and experimental foundations for enabling robots to undertake a variety of tasks with reasonable performance. For instance, the contribution in motion planning of cooperative systems made by Professor Latombe's research group at Stanford University certainly has brought significant confidence for engineers and policy makers in the automotive industry. They have developed intelligent motion planning algorithms for tasks including spot welding, car painting, inspection and measurement, where the end-effectors of a set of coordinated industrial robots reach successive goal placements given as inputs.

Additionally, due to the fast-growing health and elder-care industry, the need for multifingered robot hands with intelligence or human hand grasping and manipulation capabilities will only intensify. Note that a robot hand is a set of cooperative multiple robotic fingers. From the Stanford/ JPL robotic hand made in 1983 to the recent Shadow hand with artificial muscles, all robot hands have functional mechatronics and motion control and have a similar form to human hands. However, embedding human hand grasping abilities into such robot hands is the key to meet healthcare and manufacturing market needs and expectations of AI robotics research. Priority has to be given to the development and employment of intelligent algorithms instead of mechanical design and low level control. The author's research group has been developing algorithms which extract human hand motion cues from our everyday object- manipulation tasks using a data glove, Vicon motion capture systems and image processing techniques. Research results have shown that motion cues can be effectively employed to represent a variety of manipulation tasks. It paves the way for creating robot hands with human hand capabilities.

What would the world be without robot automation? How much further could we go by giving robots intelligence capabilities?

Honghai Liu Institute of Industrial Research, University of Portsmouth, UK