The making of a microcontroller cluster

The aim of this project was to develop a cost-effective anthropomorphic manipulator and a sensor glove, which would be capable of real-time control and haptic feedback, supported by a distributed computing microcontroller network. It investigates the design and engineering of anthropomorphic joints, force-feedback control systems and microcontroller nodes.

Ever since the late 1930s, the field of robotic engineering has gained significant attention in the scientific community. Starting off with primitive mechanical designs to perform mundane two-state tasks, engineers gradually moved on to more complex robots which were able to perform human-like activities, such as lifting objects. However, it was only in the early 1970s that the true potential of anthropomorphic robots was realized. By definition, an anthropomorphic robot is one which is built around human geometry and derived from the study of human form and action. Several robots, such as Honda’s ASIMO¹ were the result of large-scale research and development in anthropomorphic robotics over the last decade. The design of anthropomorphic manipulators is particularly challenging as it involves replicating human form, without the luxury of muscle-and-bone joints afforded to nature.

Two engineering concepts that have become inseparable from anthropomorphic robot design are those of haptic feedback and real-time distributed processing networks. The term haptic feedback refers to the process where haptic sensory information (sense of touch or force applied on a surface) is measured or calculated on a primary device and relayed to a secondary device for possible replication. Haptic feedback gives anthropomorphic manipulators the ability to handle both light-fragile objects and heavy-rigid objects. In the absence of haptic feedback, the anthropomorphic manipulator would either be too weak to handle Rigid objects or too strong for light-fragile objects.

Anthropomorphic palm with 2 fingers built out of Aluminum

The second concept of real time distributed processing networks, plays a crucial role in processing and communicating large amounts of information both within an anthropomorphic module and to external modules. The advantage of such networks is that they enable system engineers to effortlessly implement a large number of interface ports using modular programming techniques, yet at the same time, sustain high data update frequencies. Furthermore, the reduced size of microcontroller units used in anthropomorphic manipulators, allows them to be conveniently distributed in small spaces over the manipulator, with only one interconnecting 3-channel bus.

The project is divided into three distinct research areas – the ‘Sensor glove with haptic feedback’, the ‘Microcontroller clusters’ and the ‘Anthropomorphic manipulator’. Briefly stated, the sensor glove will measure the movements of a human hand, which, in turn, will be processed and communicated via the microcontroller clusters to the anthropomorphic manipulator. The anthropomorphic manipulator, in turn, will replicate the sensor glove’s movements, and at the same time provide haptic feedback to the sensor glove through the microcontroller clusters.