UM::Autonomy's news feed to keep you informed! Subscribe to our RSS feed!
|
posted Jun 10, 2011 8:21 PM by UM Autonomy
[
updated Jun 10, 2011 8:36 PM
]
The static judging portion of the competition took place today. We placed first overall out of fifteen teams. That gives us a points lead going into the qualifying round, as well as first choice of time slots for qualifying runs. Our qualifying runs are scheduled for Saturday (6/11) at 2:00 pm (EST) and Sunday (6/12) at 11:30 am (EST). This is the second straight year that we have placed first overall in static judging.
|
posted Jun 9, 2011 3:38 PM by UM Autonomy
[
updated Jun 9, 2011 9:32 PM
]
We arrived last night and got right to work. We tested our PID calibration, and all appears to be well. We took our first two runs today and logged data to analyze our system tonight. We will also be rebalancing our hulls, and reworking our servo drivers to support multiple devices. We posted some pictures of these initial tests. Check them out and wish us luck.
|
posted Jun 6, 2011 11:23 AM by UM Autonomy
We went out and bought supplies to construct a mock course and were able to test. It has been decided that our testing pool distorts the magnetic field of the compass which causes some unwanted blurring in our point cloud. We tried point cloud creation on land while moving and it looked amazing.
We now have an occupancy grid that will be used mainly for obstacle avoidance, object detection is coming along, and the camera calibration is close but still needs some adjusting.
Right now, we're getting everything ready for competition. The vans are reserved, the posters and powerpoints are being made, and things are being packed. Our first test run at competition will give us some incredibly useful data and give us a good idea of where the system is.
|
posted May 31, 2011 1:49 PM by UM Autonomy
This past weekend we were finally able to test SLAM. Our state prediction and point cloud creation is looking good. We still need to tweak some SLAM parameters to group landmarks better and try feature detection in a more competition-like environment.
We were also able to fix our issues with the IR camera driver as well as rigidly mount the IR camera. It now works well and reliably. We took the checkerboard pattern into the bright sunlight and were able to see it with the IR camera. This will allow us to calibrate the IR camera with the Hokuyo. : )
This weekend we constructed a ball grabber that works very well. There is still work to be done on controlling the RC car and mounting the APRIL tags. We'll put up a video closer to competition.
We also mounted the water cannon and waterproofed all of our open connections. With only 9 days remaining, things are starting to come together. |
posted May 26, 2011 7:16 AM by UM Autonomy
Today we wired up the FLIR PathFindIR to our system and adapted our camera driver to work with it. We still need to find the right positioning for it and lock it in place and then calibrate to the LIDAR but, it looks like it detects heat very well. This awesome IR camera from FLIR will have no trouble detecting the hot target.
Here's a picture of a team member with a soldering iron.
|
posted May 24, 2011 7:21 PM by UM Autonomy
Today we got a new larger calibration pattern printed out on poster board paper and mounted it to a big foam board. It looks and works great and will make re-calibration much easier.
To fix an issue with the camera images being a little washed out, we covered the inside and front sides of the camera box with black tape which appears to have helped.
We updated our Hokuyo, Camera, and GPS firmware so, everything should be up to date.
Finally, we were able to get in the water again and get some good logs of buoys, a ramp, and a sign with a slower moving servo speed. We got many more points on the buoys and the results are promising!
Checking on camera-LIDAR calibration tomorrow and re-calibrating if necessary, as well as working on object detection.
|
posted May 23, 2011 7:27 PM by UM Autonomy
[
updated May 23, 2011 7:30 PM
]
This weekend we were able to get the new cameras calibrated with the Hokuyo and finally got on the water to collect some data.
We got point cloud data of buoys, a sign, and a ramp. It's looking like we'll have to slow the servo down to collect more points further away. However, we did get good results from the sign and ramp data collection. Both the sign and ramp are clearly visible in the point cloud and should segment and be detected well. We changed point cloud publishing to publish only a downward or upward sweep at a time to fix issues with blurring. We still need to see what point cloud generation looks like with our new EKF.
Although the cameras are calibrated, the image is a little washed out due to reflections from inside the yellow box and incorrect lens adjustment. We'll adjust the lenses and darken the inside of the box to fix this issue.
We also were able to get the paper together and got a few good pictures of the true color point cloud and segmentation that were included. This week we'll be getting the cameras to produce a desirable image, analyzing our collected data, and getting in the water as much as possible.
Next weeks goals are to tune the PID controller, test the new EKF-SLAM system, and get good point cloud results. |
posted May 23, 2011 7:21 PM by UM Autonomy
[
updated May 23, 2011 7:32 PM
]
The new Hokuyo and power converter arrived today. We successfully installed the big new power converter - luckily having some extra room in the electronics box for it. The new Hokuyo has also been wired up and is working. It was initially failing a checksum upon initialization (consistently) in our driver but, we removed that check and it now works just fine.
We finally had the cameras and Hokuyo up and running so, we took calibration logs with our checkerboard in different positions. The cameras should be calibrated now.
|
posted May 23, 2011 7:17 PM by UM Autonomy
We heard back from our Hokuyo supplier and they are kindly shipping us a temporary replacement while they fix our Hokuyo. We should have it by Wednesday the 18th. Also, the new power converter has been ordered and is on its way.
We tried out April tags today and they work brilliantly. We still need to modify the April code to work with our system and mount three tags to the RC car. |
posted May 23, 2011 1:38 PM by UM Autonomy
[
updated May 23, 2011 7:34 PM
]
We finished the new camera box and camera box mount this past week. It is much more stable and we are able to position the cameras better.
We tried to calibrate cameras and LIDAR today but, the computers are unable to detect the Hokuyo even though it is powering on. We're looking for a replacement and contacting our supplier to figure out what we can do. It's possible that we may have strained the cables when we took the boat to the Wilson Center to fix the camera box or we may be exceeding our power converter's current limit. We'll be making sure the cables are in a safe place and we're ordering a new power converter to ensure we don't exceed the current limit.
Despite not calibrating, we did get the inputs and outputs of SLAM sorted out and discussed priorities moving forward. Hopefully we get this Hokuyo situation sorted out.
|
|
