Bioinspired Fish Robot
Autonomous Underwater Vehicles (AUVs) and Remotely Operated Vehicles (ROVs) are used for deep sea survey, wreckage exploration, oceanic mine defusal, and other aquatic scientific and military operations that pose a danger to human-operated crafts or divers. AUVs are relatively inexpensive and less labor-intensive than their ROV counterparts, however current AUVs are limited in their permissible mission duration away from shore or charging station due to the lack in battery energy density and inefficient propulsion methods. The propulsion of fish has long been studied for its high efficiency, maneuverability, and thrust performance compared to conventional propellers; however, the optimal motion of fish fin propulsors in water has not been fully characterized and thus is not ready for implementation into AUVs. To determine the optimal parameters for fish fin trajectories in a variety of flow regimes and maneuvers, the Center for Autonomous Systems and Technologies (CAST) lab at Caltech developed a biomimetic fish robot that can perform complex flapping, spinning, and oscillatory fin trajectories.
My contributions to this project ranged from the development of a control system for the fin mechanism, to a full redesign of the fin mechanism and motor.
November 2019 - Summer 2020
Having joined the Gharib Group in 2019, plans were already in motion to develop the new bioinspired fish robot. At the time, fin trajectory optimization was performed on a completed 3-DOF Spherical Parallel Manipulator (SPM) from HEPIA. Placed above an oil tank, the SPM allowed for the manipulation of a fin through complex trajectories. However, the SPM from HEPIA could not be used for experimentation of fins in co-flow regimes, as the mechanism could not be submerged and its form factor would disrupt upstream flow. The optimization of fin trajectories on the HEPIA SPM also required the complicated and bulky interfacing of multiple software programs, which made experimentation difficult and time-consuming. A new SPM was developed that could fit within a streamlined fish body, consisting of a 4-RRS redundant actuator.
The structure of the new 4-RRS actuator was constructed, and my task for Summer 2020 as a Summer Undergraduate Research Fellowships (SURF) recipient was to develop a control system and electronics suite for the robot from scratch. Utilizing an Arduino Mega and Matlab, I developed software that transmits a trajectory using a serial communications protocol that I built myself, stores the trajectory data onto the Arduino, and performs the trajectory on the mechanism.
I solved the inverse kinematics of the mechanism both numerically and analytically. As a closed-chain redundant actuator, I artificially constrained the height of the end-effector via software, which effectively limits the fin mechanism to three rotational degree of freedoms. Thus, only the desired actuator joint angles were sent to the Arduino, minimizing memory usage and processing time.
My work during SURF 2020 culminated with the successful operation of flapping and spinning trajectories for the fin mechanism. Water tunnel experimentation was not possible during this period, as work on the fin mechanism took place completely remote. However, during my testing of the mechanism I noticed some design flaws of the fin mechanism. First and foremost, I noticed that the motor meant for axial rotation would not provide enough torque for operation in water. I also observed a mechanical offset of the motor shafts and linkages, which may lead to inefficiencies and unintended motion. I also found that the IMU and central motor controller were not sufficient for the desired motion. Fixing these problems would be the focus of my next work on the project.
Fall 2020 - Summer 2021
After having successfully developed a controls system, my first goal was to source a new motor for the axial rotation. This axial rotation allows the fin to act as a normal boat propeller, combining both human technology and biologically-inspired propulsion. The motor had to fit a wide variety of criteria, such as power requirements, size constraints, weight constraints, cost, etc.. The search for a motor was also exacerbated by supply-chain shortages, as many motors from various companies were not in stock and would take months to arrive.
The primary concern was the power requirements for the motor and finding a motor capable of providing the necessary power while minimizing its size. The performance requirements were calculated with the maximum torque and speed desired for experimentation, including a factor of safety. This analysis resulted in a necessary torque of 3.4Nm at 180rpm, leading to a power requirement of about 70W.
Since the motor and mechanism must fit within the fish body, increasing the motor size and shaft size would have major effects on the achievable tilt angle of the fin. Therefore I performed detailed analyses on the range of motion given the dimensions of a motor. This analysis showed that the motor had to be either short and flat (Pancake motor), or tall and skinny.
I then focused on the design of the new motor mount and fin mechanism structure.
To integrate the motor into the existing assembly and fix design flaws of the previous mechanism, seven parts were designed including mounting brackets, bearing supports, and linkages. Designs prioritized minimizing the size, weight, machining time, and cost of the assembly. Designs were implemented in CAD and given the correct weight characteristics. Stress analyses were performed in Solidworks for critical components such as linkages based upon the maximum operating limits of the motor. Critical components were designed with a large factor of safety, and all parts are comprised of 6061 Aluminum.
To ensure ease of machinability and low cost, designs were discussed with the
GALCIT machine shop and iterated upon. Once the designs were finalized, detailed drawings
were drafted and sent to the machine shop along with the purchased necessary materials. The design also included a mounting location for the IMU sensor to be used in data collection.
The new motor and fin mechanism has been assembled, and will be used for experimentation in a water tank.