We further troubleshooted our new program for the TorsoBot. Initially, its legs were not moving in the way that we wanted them too. Rather than go straight forward, they would go back first then forward. We fixed this problem by defining the angles within the while loop rather than using a function. Our next problem was that the leg on servo 2 was not on the ground at the same time as servo 0. In order to solve this problem, we added small, freely swinging triangular blocks to the legs. This did not work. The robot's wheels didn't get enough force to begin spinning and the triangular blocks did not return to their initial position after the push off. My theory is that if we could make the legs fall a little bit further forward, it would give the blocks time to return to their position. however, this is impossible for me to do because I would need an angle greater than 160 for servo 2. The maximum angle that servo 2 can travel is 160.
In order to learn more about how the triangle blocks affected the motion of the robot, I ran a series of tests changing only the starting position of the block and the desired speed of the robot. There were a few successes and a few failures as usual.
On the second day of the week, we wrapped a rubber band from the tail, around the torso, to the head and the robot actually worked! The question is whether or not this is "cheating". This robot is supposed to represent a human walk, the rubber band makes it seem like a human's ankle is attached to their head (which is not true).
Dr. Adamczyk came and talked to us about it and redirected us to the original focus of the project. The point was not to make long, slow strides, but to make quick, short push offs. We planned on trying a different leg position and decreasing the difference between the start and end angles to make a shorter push off. However, while we detached the rubber band, one of the pieces connecting the tail to the axel snapped. Before we could do any more testing on leg positions, we would have to laser cut and attach a new piece.
This did give us time to test the TorsoBot by itself, without the legs. Our lab needed 50 data points for these tests with different torso angles and desired speeds. However, when trying to remove the broken pieces of the tail, we discovered that the holes on the rimless wheels were not in the same spot. This held back our testing because we had to find a way to fix the wheels before we could begin. In addition, the two of the spokes broke off while we were trying to fix the robot. For this reason, one of the other interns designed a new set of wheels with supports in between the spokes. This further setback our project as we had to wait to do any testing until the wheels were made.
On a good note, that day, Dr. Adamczyk dropped off a new raspberry pi for the robot arm (another one of our projects). For the rest of the day, we set up the pi and plugged in the servos. This raspberry pi worked for approximately twenty minutes before it shut down and prevented us from connecting to it.
The next day, we continued to try and fix the pi. Thankfully, the SD card was not faulty so we simply switched over to a different pi. After a few hours of set up, we were able to connect easily and begin work on the program. Dr. Adamczyk provided us with a program for the robot, but this program did not seem to work. We received multiple errors and the servos remained still. The most common error was "Error accessing 0x40: check your I2C address". Despite multiple troubleshooting attempts, the error seemed to appear and disappear randomly. In addition, we still couldn't get the servos to move. It turns out that not only was our raspberry pi from the previous day fried, but also the servo hat we had been using. We simply used a different raspberry pi and servo hat and the system worked fine. It was very exciting to see the robot arm move. At that point, we only had a program that moved each servo individually using a digital slider. We were unsure what to do next with the robotic arm and so we put this project on hold until we could confer with Dr. Adamczyk.
On the last day of the week, we attempted to use the motion capture cameras to collect data on the TorsoBot. Although we still did not have the other reinforced wheel, we were able to run some basic tests without the tail.
Comments