Control Systems Design of a Soft Bodied Endoscope
Undergraduate Research Student: Steven Warren
Advisor: Prof. Barry Trimmer
Co-Advisor: Prof. Valencia Joyner
Funding: Tufts Biomedical Research Experiences for Engineering Majors (BREEM)

This project is part of a multi-disciplinary research initiative at Tufts to investigate the science and engineering of a new class of soft-bodied biomimetic robots, incorporating biomaterials, neuromechanical controllers, and evolutionary concepts. The design of conventional robots places emphasis on achieving the abstract objective of movement itself. While movement may be achieved in this fashion, approaching the construction from a biomimetic standpoint allows the designer to model the robot based on a known working model – a living organism.

This study highlights the biomimetic design considerations of a soft bodied robot relative to the emulation of Manduca sexta motor control. The replication of Manduca kinematics within a robot would allow for a maximal degree of freedom in confined spaces. Unlike other soft bodied animals, the Manduca has the ability to climb; an advantageous characteristic which allows additional freedom of movement. The benefit of these properties could be seen in endoscopy; while today endoscopy relies largely on rigid devices, the freedom of movement found in a soft bodied robot may improve endoscopic diagnostics.

The control system for the soft bodied robot will be based on the concept of a central pattern generator. A central pattern generator is the primary means by which animal movement occurs. A rhythmic pattern is distributed throughout the motor neuron network of the animal and movement is achieved. These patterns may be changed accordingly by the brain depending on the circumstance or desired movement. The actual patterns have been recorded as action potentials. The potentials have set duration that varies dynamically in frequency and duty cycle to control the degree of actuation. The control system for the robot will implement a means to mimic this behavior through both direct interface with Labview ™ and an onboard PIC microcontroller.

Since movement patterns for the robot are not yet known, the control systems must be flexible enough to run under a genetic algorithm. A genetic algorithm is a computer algorithm that naturally changes variables over generations of trials until a proper goal has been achieved – in this case movement similar to that of the Manduca. One major advantage of this approach is the automation of pattern discovery.

A board level interface circuit was developed to apply concepts and genetic algorithms derived by Trimmer’s Lab and collaborators in Computer Science to control actuation and movement. Robot actuation is enabled by applying high-power frequency modulated signals to an electrically heated shape memory alloy (SMA) material. This project will continue during the academic year with an undergraduate and graduate research assistant investigating the use of 180nm devices and low-power circuit design techniques to scale/integrate the entire control system and interface circuit on a single microelectronic integrated circuit.

Keywords: CPG (Central Pattern Generator), Endoscope, Soft Body Robot, Biomimetic

Back to Research