Videos of Old Robots That I Built or Worked On
LSU RoboKitty - Early Walking Experiment
This video shows the RoboKitty, a project that I worked on at LSU in the early 2000s, while working in the old Robotics Research Lab (RRL) in Coates Hall at the LSU Baton Rouge campus.
At Dr. S. S. Iyengar's bidding, I designed and built this “cat” robot using RC servos and a small control board. My intention was for it to learn how to walk, and my idea to accomplish this was to use a genetic algorithm.
Simulation and physical equipment are different in many ways; the time and physical toll on the robot being the key differences. After a few hundred robot walks, breeding, ranking, etc., that became apparent. This video shows some of the later results.
LSU RoboKitty - Later Walking Experiments
The videos show the RoboKitty walking using gait control and leg movement software. The system worked using LabVIEW software generating control signals, which were sent out the RS232 port of the computer to an SSC-32 servo controller.
This is the same robot and servo control system as used in the earlier experiments. However, the new software control was much quicker and less detrimental to the robot's locomotion hardware. In the second video, you can see Mitzy, a Siamese cat, coming to check out the robot.
Associated Papers and Presentations
Open RoboKitty Associated Paper
Open RoboKitty PowerPoint Given at University of Southern Mississippi, October 2000
Naval Research Lab, Stennis, MS - Robot Formation Maneuvering Experiments
These videos show the first stages of robot formation maneuvering experiments conducted at NRL in the UNCL by myself, Dr. Brian Bourgeois, Marvin Roe, and Don Brandon. We also had excellent help from Josh Miceli, Dustin Marmalich, John “Kerry” Herrington, and others.
Our goal was to use relative positioning to widen the sensor footprint of autonomous robots. There was to be a lead robot, with followers whose purpose was to extend the sensing capability of the leader. The end goal was to use this system underwater, thus the acoustic tracking.
The robots detect each other by sensing an acoustic chirp and following it. Each robot followed a chirp in an assigned frequency band. In the lab environment, severe echoing made this difficult, but the chirping worked.
This video shows testing of the next stage of the formation maneuvering project: water-based vehicles. Model boats were selected to be the water “robots.” Much work was put into this, including fitting the robots with PC-based computer systems and working with hydrophones.
Southeastern Louisiana University - Robot Hockey
Although seemingly unrelated, I was still focused on learning limb control. The hockey system was limb control, but with no balance; an easier problem for sure. However, depending on how quickly the puck is moving, success can be elusive.
Several students worked on this project. Dan Reid and Robert Chaney made large contributions to the project. Matt Lakin's hard work was key to its success; he is in the video on the left. The video on the right shows some improvement in the tracking and control systems.
Southeastern Louisiana University - Large Actuator Test Bed (LATB)
The Large Actuator Test Bed project was another study in limb control. This time the goal was to get experience with low-cost linear actuators and check their viability for use in a quadruped robot.
The system was controlled using two laptop computers: one controlling the robot walking functions, and another using Virtual Network Computing (VNC), so that the robot could house a PC for its computational needs while an operator monitored and controlled movement.
The linear actuators were powered by Victor motor speed controllers, using PWM signals from a control board taking signals from the onboard PC. The control system relied on process synchronization supported by the Bakery algorithm. Everything worked well until the primary computer's battery failed and the system had to be ported to a different computer.
Southeastern Louisiana University - HCR (Headless Cat Robot)
The HCR project was a modernized version of the RoboKitty from years before. This time Dynamixel servos were used instead of RC servos.
The Dynamixel units house small microprocessors that allow them to be daisy chained together, rather than having an individual wire going to each servo. Another advantage is that engineering information from the servo can be read, providing feedback such as servo position and stall conditions.
The system was controlled using an Arduino board set up for controlling the Dynamixel servos. The robot's body and legs were constructed using Bioloid robot brackets and other parts.
Was this more successful than the original? In some ways yes, in others, not so much. The servos and control system were a step up, but the Bioloid parts were heavy and not really well suited for a cat-like robot.