This paper describes a technique for improving dynamic performance of robots by recursively modifying the mechanical structure through prototyping and experimentation. In each recursion, a prototype robot is tested and evaluated experimentally, an incremental change in structure design is determined based on the analysis of the experimental data, and the mechanical structure is physically modified so as to drive the plant dynamics towards a desired response. To expedite the iterative process, (i) a rapid prototyping technique using photo-polymerization is developed, (ii) a gradient descent method is applied to recursively determine an optimal design, and (iii) all the design, prototyping, and experimentation processes are integrated and carried out under computer control. The optimal, incremental change of design is determined by using a sensitivity Jacobian. The sensitivity Jacobian is initially obtained numerically using a finite element model. Further, the sensitivity Jacobian is corrected and updated recursively with experimental data after each iteration. A proof-of-concept demonstration system is built and tested for the development of a simple single link robot. The arm structure made of plastic is reinforced recursively by coating it with a photo-acrylate plastic with an optimal thickness distribution so that the frequency response of the structure can be improved toward a desired reference model.

ADINA, Software for Finite Element Analysis, 71 Elton Av., Watertown MA 02172.
Deitz, D., 1991, “Stereolithography Automates Prototyping,” Mechanical Engineering, Apr.
Goodwin, G. C., and Sin, K. S., 1984, Adaptive Filtering Prediction and Control, Prentice Hall.
Jordan, M. I., and Rumelhart, D. E., 1992, Forward Models: Supervised Learning with a Distal Teacher, Cognitive Science, 16.
Narendra, K. S., and Annaswamy, A. M., 1989, Stable Adaptive Systems, Prentice-Hall.
Sachs, E., Cima, M., Brancazio, D., et al., 1990, “Three Dimensional Printing. Rapid Fabrication of Molds for Casting,” Advances in Integrated Product Design, ASME.
Sanner, R. M., and Slotine, J. J. E., 1992, “Gaussian Networks for Direct Adaptive Control,” IEEE Trans. of Neural Networks, Vol. 3, No. 6.
Weiss, L., Prince, S. D., and Adams, D., 1992, “Thermal Spray Shape Deposition,” ASM Journal of Thermal Spray, Vol. 3.
Yang, B.-H., and Asada, H., 1993, “A New Approach to Adaptive Reinforcement Learning,” Proc. of 1993 Int. Joint Conference on Neural Networks.
This content is only available via PDF.
You do not currently have access to this content.