Team 31 CART Update (01/31/2020)
Thanks for reading our blog! If this is your first time seeing on our blog, you might want to read our first blog post.
We’ve hit the ground running this semester. So far, we’ve installed an upgrade to the braking system to allow an additional (more rapid) braking mode to be activated in the event of an emergency. This is a necessary safety feature before we can begin any autonomous driving tests as the pre-existing braking was insufficient for emergency situations. We have also replaced the pre-existing remote controller with a standard COTS 3-channel transmitter/receiver and tuned the motor controllers to respond to the new remote controller. We plan to verify/validate the remote control functionality of the CART within the week.
We’ve also done a bit of upkeep for the CART’s autonomous infrastructure. We improved the reliability of the CART’s control system by properly soldering and shrink wrapping many connections made by the previous team. We also re-designed and re-printed the plastic bracket that mounts the steering motor onto the steering column. The old mount cracked down the middle and was slipping, so we beefed up the mount (where thickness wasn’t critical for gear teeth alignment) and adjusted print settings to increase the strength of the part significantly.
We faced some challenges while accomplishing the tasks listed above.The initial pneumatic solenoid (for the braking system) received from AndyMark was defective, and we had to return the part. Luckily, AndyMark acted promptly and replaced the defective part within three days. However, when testing the brake system, we realized that there was an issue caused by the addition of the second solenoid in which the air released by one valve vented out the other valve’s open return air port. This was remedied with a cheap one-way valve installed between the solenoids. After replacing the faulty solenoid and fixing the design issue, we tested the brake compressor and blew the 20A fuse for our 15A compressor. We suspect that the unusually high current draw was due to the compressor motor heating up during the initial pressurization from 0 to 150 psi. The hot motor lost efficiency and began drawing more current. We will adjust our procedures for this initial pressurization by shortening the duty cycle to allow the motor to stay cool under the higher loads.
We have also experienced some unexpected delays. We anticipated making some modifications to the CART’s cable management, but we did not expect the number of damaged/loose connections that needed to be fixed. The cable modifications stretched over a week. The wiring system created by the previous team was messy and unwieldy. Ideally, we would like to replace this system with a cleaner/more intuitive setup, but since the wiring system is functional, this goal is not a high priority.
In our initial project proposal, we planned to use a LIDAR as our primary perception sensor. Unfortunately, due to setbacks in acquiring the funds to purchase a LIDAR unit, we haven’t selected/mounted a LIDAR unit on the CART. At the moment, we are still stuck with the problem of selecting an affordable 3D perception system. Without a robust perception system, the CART can’t navigate safely.
We currently have enough funds to purchase a mid-quality, 2D LIDAR unit. We delayed in purchasing a 2D unit because we hoped to raise enough funds for a more robust 3D unit. Preliminary tests using a 2D LIDAR produced unimpressive (poor quality, low volume) results. To navigate safely, we need high quality data, and lots of it.
Further compounding our fundraising issues is a recent announcement by Velodyne in which they promised to release a mid-quality, solid state 3D LIDAR unit for $100. This unit (called the Velabit) promises specs similar to the Livox and LeddarTech (both commercially available) solid state units at a much lower price point. We had planned to purchase one of the available solid state units for ~$1000. The potential release of the Velabit makes the purchase of either the Livox or LeddarTech LIDAR a bad financial proposal because would essentially be squandering $900.
We are currently deliberating on our next steps, but we are exploring the possibility of combining 1D laser rangefinders (a performant and free replacement of our ultrasonic sensor) with a depth camera. Both sensors are freely available (thanks to professors from the University of Houston). This solution would be less desirable, but it would allow for the option of autonomous driving obstacle detection. We are also exploring the option to borrow a 3d LiDAR (pending the release of the Velabit) and other fundraising possibilities.
In parallel, we intend to finish configuring the Pixhawk in Mission Planner/QGroundControl (a desktop software application) and integrate it into the remote driving system. We anticipate some trivial wiring issues to solve in this process and hope to have it completed by February 7th. Once we have the Pixhawk configured and integrated onto the CART we will start tuning the driving model by running basic GPS navigation tests. Though none of us have used the Pixhawk in this context before, we are confident, from previous experience with the Pixhawk, that it will be a straightforward process. These tests will be very safe, as we are able to switch between the RC transmitter, autonomous, and manual control very easily with the Pixhawk, and a driver will be monitoring from behind the wheel at all times.
Once we finish configuring and begin tuning the system, we’ll need to permanently mount the pixhawk and its accessories. The challenge is that our mount needs to be secure and isolated, but must paradoxically be easy to access and simple to detach. Additionally, the mount must also sufficiently dampen any vibrations (caused by driving) as to not interfere with the accelerometers on the board. We will be developing this mount in parallel with everything else, and it should only take a day or two to print. Since the Pixhawk is light (less than 100g) and generates no force, this mount will not need to be analyzed with a finite element analysis.
We will also be developing the odometry system and installing it onto the CART. This system will just be some permanent magnets mounted to the rim of a wheel and a Hall effect sensor mounted to the stationary part of the hub. This should give us a fairly simple signal that we will need to write some software to convert to a speed reading and then pipe that reading into the navigation computer. This software should be straightforward to write.
Progress since our last post
It has been a while since we last posted. Since our last post, we’ve ordered and received many of our components. We had planned to work on the CART over the break, but due to delays in the procurement process, many of our parts didn’t arrive until the end of the break. Unfortunately, this meant that we were unable to get much work done during the winter break.We’ve hit the ground running this semester. So far, we’ve installed an upgrade to the braking system to allow an additional (more rapid) braking mode to be activated in the event of an emergency. This is a necessary safety feature before we can begin any autonomous driving tests as the pre-existing braking was insufficient for emergency situations. We have also replaced the pre-existing remote controller with a standard COTS 3-channel transmitter/receiver and tuned the motor controllers to respond to the new remote controller. We plan to verify/validate the remote control functionality of the CART within the week.
We’ve also done a bit of upkeep for the CART’s autonomous infrastructure. We improved the reliability of the CART’s control system by properly soldering and shrink wrapping many connections made by the previous team. We also re-designed and re-printed the plastic bracket that mounts the steering motor onto the steering column. The old mount cracked down the middle and was slipping, so we beefed up the mount (where thickness wasn’t critical for gear teeth alignment) and adjusted print settings to increase the strength of the part significantly.
 |
| Broken and old steering mount (left) vs new and stronger mount (right) |
We faced some challenges while accomplishing the tasks listed above.The initial pneumatic solenoid (for the braking system) received from AndyMark was defective, and we had to return the part. Luckily, AndyMark acted promptly and replaced the defective part within three days. However, when testing the brake system, we realized that there was an issue caused by the addition of the second solenoid in which the air released by one valve vented out the other valve’s open return air port. This was remedied with a cheap one-way valve installed between the solenoids. After replacing the faulty solenoid and fixing the design issue, we tested the brake compressor and blew the 20A fuse for our 15A compressor. We suspect that the unusually high current draw was due to the compressor motor heating up during the initial pressurization from 0 to 150 psi. The hot motor lost efficiency and began drawing more current. We will adjust our procedures for this initial pressurization by shortening the duty cycle to allow the motor to stay cool under the higher loads.
We have also experienced some unexpected delays. We anticipated making some modifications to the CART’s cable management, but we did not expect the number of damaged/loose connections that needed to be fixed. The cable modifications stretched over a week. The wiring system created by the previous team was messy and unwieldy. Ideally, we would like to replace this system with a cleaner/more intuitive setup, but since the wiring system is functional, this goal is not a high priority.
 |
| Wiring system created by previous team which controls the actuators on the CART |
A note on LIDAR
In our initial project proposal, we planned to use a LIDAR as our primary perception sensor. Unfortunately, due to setbacks in acquiring the funds to purchase a LIDAR unit, we haven’t selected/mounted a LIDAR unit on the CART. At the moment, we are still stuck with the problem of selecting an affordable 3D perception system. Without a robust perception system, the CART can’t navigate safely.We currently have enough funds to purchase a mid-quality, 2D LIDAR unit. We delayed in purchasing a 2D unit because we hoped to raise enough funds for a more robust 3D unit. Preliminary tests using a 2D LIDAR produced unimpressive (poor quality, low volume) results. To navigate safely, we need high quality data, and lots of it.
Further compounding our fundraising issues is a recent announcement by Velodyne in which they promised to release a mid-quality, solid state 3D LIDAR unit for $100. This unit (called the Velabit) promises specs similar to the Livox and LeddarTech (both commercially available) solid state units at a much lower price point. We had planned to purchase one of the available solid state units for ~$1000. The potential release of the Velabit makes the purchase of either the Livox or LeddarTech LIDAR a bad financial proposal because would essentially be squandering $900.
We are currently deliberating on our next steps, but we are exploring the possibility of combining 1D laser rangefinders (a performant and free replacement of our ultrasonic sensor) with a depth camera. Both sensors are freely available (thanks to professors from the University of Houston). This solution would be less desirable, but it would allow for the option of autonomous driving obstacle detection. We are also exploring the option to borrow a 3d LiDAR (pending the release of the Velabit) and other fundraising possibilities.
Planned work for the near future (January 31 - February 14)
Besides sourcing our main perception sensor and some other small components (such as our ultrasonic sensor), we’ll be running a verification test on the remote driving system. This will verify the reliability of the driving hardware onboard and simplify troubleshooting down the road.In parallel, we intend to finish configuring the Pixhawk in Mission Planner/QGroundControl (a desktop software application) and integrate it into the remote driving system. We anticipate some trivial wiring issues to solve in this process and hope to have it completed by February 7th. Once we have the Pixhawk configured and integrated onto the CART we will start tuning the driving model by running basic GPS navigation tests. Though none of us have used the Pixhawk in this context before, we are confident, from previous experience with the Pixhawk, that it will be a straightforward process. These tests will be very safe, as we are able to switch between the RC transmitter, autonomous, and manual control very easily with the Pixhawk, and a driver will be monitoring from behind the wheel at all times.
Once we finish configuring and begin tuning the system, we’ll need to permanently mount the pixhawk and its accessories. The challenge is that our mount needs to be secure and isolated, but must paradoxically be easy to access and simple to detach. Additionally, the mount must also sufficiently dampen any vibrations (caused by driving) as to not interfere with the accelerometers on the board. We will be developing this mount in parallel with everything else, and it should only take a day or two to print. Since the Pixhawk is light (less than 100g) and generates no force, this mount will not need to be analyzed with a finite element analysis.
We will also be developing the odometry system and installing it onto the CART. This system will just be some permanent magnets mounted to the rim of a wheel and a Hall effect sensor mounted to the stationary part of the hub. This should give us a fairly simple signal that we will need to write some software to convert to a speed reading and then pipe that reading into the navigation computer. This software should be straightforward to write.
Comments
Post a Comment