Past and Future Work on the MIT RoboTuna

Past Work

The overall experimental strategy consisted of three phases. The first phase constituted flight checking the actuators, sensors and control software during simple body motions while the RoboTuna was parked at the dock. This phase was primarily to determine if each of the six body joints moved to the correct location at the corresct speed when commanded by the control software as well as if each joint's force and dispalcement tranducers produced accurate and repeatable readings over the full range of possible joint motions.

The second phase consited of a set of basic "full swimming" test. The RoboTuna basically swims by passing a wave variable amplitude down its body from nose to tail. The Tuna's body motion control software creates this wave based on a set of 7 experimental parameters. In order to verify full system operation, repeatability, and stability, an array of tests were performed in which the Tuna was required to swim the Tank at normal flight speed.

In a typical test, a specific set of swimming parameters were loaded into the body controller. Using the body wave prescribed by these parameters, the oscillation Tuna swam the full length of the testing tank. Durring this run the Tuna's high speed data collection system simultaneously captured, stored and archived all relevant sensor data. This captured force and displacement data was then post processed, generating a quantitative measure of simming performance for that specific set of parameters. The aggregae sum of the full array of tests conducted during this phase of the experimental program gives a clear picture of the overall robustness, capability and repeatability of the apparatus.

The third and final phase is a search for the optimum swimming performance obtainable within the physical limits imposed by the design of the RoboTuna and the length of the existing testing tank. The current analytical intractability of the fluid dynamics of this problem indicated that the most pragmatic way to proceed would be to optimize the body wave controller experimentally. In simple terms, given the seven parameters which control the swimming body wave, this can be thought of as an experimental search through seven dimentional space. This large number of dimensions quickly creates a massive logistics problem (about 282,475,249 combinations of parameters).

Given that it takes approximately 5 minutes to make a single experimental run down the tank, it would take a time frame in the order of millions of years to perform a blind search through all the combinatorial possiblities in the persuit of an optimum (it is no coincidence that this is about the same amount of time it took for the biological tuna to evolve to its present form). Obviously a more efficient search mechanism is needed, in orger to find the optimum before either time ran out or the apparatus failed mechanically. After a survey of many existing multidimensional space search techniques, a robust, seft-optimizing system based on a Genetic Algorithm was developed.

Current Work/Acheivements

The experimental results of tests clearly demonstrate that RoboTuna duplicates Gray's paradox (i.e. the drag of the swimming fish RoboTuna apperars to be less than the drag on the straight RoboTuna), but does so with unarguably "known" mechanical muscles. These experimental results, at least for the parameters tested, support Gray's claim that differences in marine/terrestrial muscle power are not the answer, but do not go so far as to explain what the solution to the paradox is. However, it does lend strong credence to the possiblity that some form of unconventional, highly beneficial hydrodynamic mechanism exists, which reduces drag in fish-like propulsion.

Based on this background, consider the fundamental questions raised, by the biological data, and how the Tuna's experimental results address them:

1) Can the flow past an undulating body propelled by an oscillating foil be "tuned" such that the body's drag is reduced and its thrust is enhanced in a beneficial way?

Yes, clearly it can be, as both Gray's paradox suggests and the results of RoboTuna's experiments conclusively show, this flow can be altered by the correct body-wave/tail-foil motion to use the hydrodynamics in a benificial way.

2) What are the parameters which control this tuning?

In the case of the RoboTuna the parameters which control this tuning are the set of traveling body-wave/tail-foil control parameters given by: *the forward speed *the tail fin's maximum angle of attack *the wavelength of the travelling body wave *linear amplitude of the body wave *a set of fluid dynamics parameters. These may not be the only or the optimal set of such parameters. As with coordinate or modal systems, there is probably an infinite variety of ways to express the same charateristics, but this particular set is eplicitly tied to body dynamics, in such a way as to be easy to observe, to measure, and to comprehend.

3) What is the maximum benifit that can be achieved?

Gray's paradox implies that a seven-fold reduction in drag be acheived. The RoboTuna only experimentally reduced its drag by about half. Obviously Mother Nature is the better engineer. For the sake of argument, assuming the RoboTuna has a purely internal mechanical efficiency in the range of 90%, by extension, it can be claimed that its apparent reduction in drag is in fact probably in the range of 60%, but obviously, there may still be a way to go.

4) Can a man-made (non-biological) system successfully exploit this phenomenon?

Yes, as the RoboTuna clearly demonstrates, it can. But a more appropriate question based on the Tuna's results is now; What level of performance can be acheived by a man-made system, redesigned based on the information collected during this phase of the program? These results also raise a host of intriguing new questions such as; What happens at higher speed?, What is actually going on in the flow about and behind the body?, and Can this be replicated in a free swimming fish?

Future Work

In terms of possible dirrection of future work, three main areas of research stickk out. First, of course, come the questions such as; What exactly is going on in the flaw about and behind the RoboTuna as it swims? and What is the effect of variations in the body-wave/tail-foil parameters on this flow? Speculating on the answers to these questions is well beyond this discussion, but using flow visualization techniques is clearly the next logical step to take. In addition, the development of an accurate (numerical) model of the flow would be invaluable.

Secondly, inspecion of the power data revealed the importance of incoporating some form of Tuned-Harmonic-Drive into the design of any future oscillation system. If the main structural element of future fish can be carefully designed such that its primary model shape corresponds to the wave form required for swimming, it is possible that much better levels of swimming perfomance can be achieved. Rough calculations indicate that installing a carefully designed THD can reduce the power requirements of a direct drive by half. If this THD can pick up on and be excited by the dynamics of the fish's wake even greater levels of performance may occur.

Thirdly, the RoboTuna is designed to really only do one thing, swim in a straight line well. Biological fish add to this whole host of amazing maneuvers. They can potentially accelerate at levels exceeding 10 g's and turn in less than half a body length at full speed. The most common questions asked by visitors to the Testing Tank about the RoboTuna is: When are you going to take it off the sled? and When will it be free swimming? The level of complexity involved in a free swimming RoboFish is many times greater than in the RoboTuna, but is the next logical step in the progression. Even now graduate student John Kumph is in the process of building and tuning an autonomous RoboPike.

A free-swimming RoboFish could be used to explore a wide range of efficiency, fast-starting, sharp turning and three-dimensional maneuvering issues as well as act as a testbed for a whole host of new marine sensors, actuators and controller technologies. The amount of fundamental new information that could be collected by such a free swimming system is immense. Are there as of yet unknown, "benificial hydrodynamics" that biological fish exploit to perform their amazing maneuvers? We believe that the understanding, technology and design expertise is there to build a free-swimming RoboFish. John Kumph has taken it uppon himself to try and find out.