Team 31 CART Update (02/14/2020)
Thanks for reading our blog! If this is your first time seeing on our blog, you might want to start by reading our first blog post.
We’ve also considered using other depth cameras, but we are reluctant to switch because this would involve purchasing a new depth camera. As long as the Intel RealSense works and its existing/supported codebase is sufficient to use in our software, there is no reason to switch to an expensive albeit more capable camera.
We also tested the remote control capabilities of the CART. We verified that the new emergency braking system is able to stop the golf cart within 25 feet. Unfortunately, this was not a full test because the metal wire connecting the gas pedal to its motor fatigued and broke away during motor tuning. We were unable to replace the wire before completing our test. However, this does not significantly change the results of our test. We have since replaced this wire with a stronger cable. We also made small adjustments to the wire’s connections to lower stress concentrations and prevent the occurrence of future fatigue failure modes.
While testing the remote control system, we discovered that the CART is unable to turn the wheels (and the steering column) when it is parked. While moving, the CART has no problem turning the steering column, but the steering motor does not have enough power to turn the wheels at a dead stop. If attempted, the motor heats up and starts to smoke (and burn through the motor’s brushes). To prevent permanent damage to the motor, we plan to create an exception (in the autonomous control code) to prevent the cart from trying to turn the wheels while it’s stopped. This slightly reduces the maneuverability of the CART, but we believe the impact on navigation to be relatively minor.
Additionally, we plan to conduct more tests on the Pixhawk. We plan to implement the GPS waypoint navigation feature as this is the basis for the rest of our platform. In the future, we will refine this to be less crude, but GPS waypoint navigation is generally easy to get working out-of-the-box. We are believers in the rapid prototype and testing mindset, where we constantly iterate and improve. This is a good approach for this project especially, because even though we have plans, some things could be much more difficult to implement than anticipated (e.g. not hitting objects at 10 mph, so we might scale back to 5 mph).
On the less technical side of things, we’d like to do a few other things. First, we want to find the laptop that the previous team used for software development on the CART. Apparently, this laptop was purchased for the CART project, but we currently do not know where it is. We have reached out to the previous owners of the CART project and hope to hear from them soon. This is important because although we plan on having an Nvidia Jetson on board, it may prove insufficient for our computing needs, and a more computationally capable laptop would be good backup.
Progress since our last post
Since the last post, we’ve fully pivoted away from using 3D LIDAR. Feel free to read our last blog post for more information, but the key point is that we don’t have enough money to buy a 3D LIDAR (but we hope future teams will be able to purchase and use the Velabit when it is released later this year). Instead, we’ll utilize depth cameras and high precision, 1-D laser scanners. We have a free depth camera (an Intel RealSense, courtesy of IEEE Makerspace) and a couple of loaner laser scanners (courtesy of the NCALM). We were able to test the RealSense earlier this week. After reconfiguring the camera for a couple of hours, we were able to get indoor depth readings for distances of up to 23 feet, (~7m) which is great, but below our goal of 10m. On a bright but cloudy day, we were able to get depth readings of up to 40 feet (~12m). While the results are promising, more testing needs to be done in order to better understand the capabilities of the RealSense |
| Depth output from the Intel RealSense, from the outdoor test |
We’ve also considered using other depth cameras, but we are reluctant to switch because this would involve purchasing a new depth camera. As long as the Intel RealSense works and its existing/supported codebase is sufficient to use in our software, there is no reason to switch to an expensive albeit more capable camera.
We also tested the remote control capabilities of the CART. We verified that the new emergency braking system is able to stop the golf cart within 25 feet. Unfortunately, this was not a full test because the metal wire connecting the gas pedal to its motor fatigued and broke away during motor tuning. We were unable to replace the wire before completing our test. However, this does not significantly change the results of our test. We have since replaced this wire with a stronger cable. We also made small adjustments to the wire’s connections to lower stress concentrations and prevent the occurrence of future fatigue failure modes.
 |
| The new accelerator pedal wire, installed on the CART |
While testing the remote control system, we discovered that the CART is unable to turn the wheels (and the steering column) when it is parked. While moving, the CART has no problem turning the steering column, but the steering motor does not have enough power to turn the wheels at a dead stop. If attempted, the motor heats up and starts to smoke (and burn through the motor’s brushes). To prevent permanent damage to the motor, we plan to create an exception (in the autonomous control code) to prevent the cart from trying to turn the wheels while it’s stopped. This slightly reduces the maneuverability of the CART, but we believe the impact on navigation to be relatively minor.
Planned work for the near future (February 14th - February 28th)
We plan to devise a method to attach the 1D laser rangefinders to the CART. We plan to attach the RealSense to the CART using the manufacturer provided mount. We have temporarily attached the odometry prototype but we are delaying tests until we obtain more magnets of suitable size and strength.Additionally, we plan to conduct more tests on the Pixhawk. We plan to implement the GPS waypoint navigation feature as this is the basis for the rest of our platform. In the future, we will refine this to be less crude, but GPS waypoint navigation is generally easy to get working out-of-the-box. We are believers in the rapid prototype and testing mindset, where we constantly iterate and improve. This is a good approach for this project especially, because even though we have plans, some things could be much more difficult to implement than anticipated (e.g. not hitting objects at 10 mph, so we might scale back to 5 mph).
On the less technical side of things, we’d like to do a few other things. First, we want to find the laptop that the previous team used for software development on the CART. Apparently, this laptop was purchased for the CART project, but we currently do not know where it is. We have reached out to the previous owners of the CART project and hope to hear from them soon. This is important because although we plan on having an Nvidia Jetson on board, it may prove insufficient for our computing needs, and a more computationally capable laptop would be good backup.
Comments
Post a Comment