I finished creating the brute-force SLAM algorithm for the lidar project. Simply put, the algorithm tests all possible movement, calculates the offsets of all these movements and then it uses the calculated movement with the least offset. The algorithm seems to be pretty reliable. It was made in OpenCL. The algorithm must run on a GPU, as a CPU isn’t powerful enough to handle all the calculations. OpenCL enables this and also opens the path to heterogenous computing (computing on more than 1 core.) A high-end graphics card is not required, my Intel HD 4000 integrated in my i5-3320m can easily run it.
I already created the mouse and gyro sensor SLAM for the lidar project, but I will write another program for it. This program will involve using a brute force algorithm and using a GPU. An article explaining this will be published shortly.
Please fill out the following form: Traffic Form
I have faced many problems in the SLAM algorithm based on the lidar readings, so I’m putting a mouse on the robot and use that mouse for optical tracking. The software code is already completed, it’s the github of the project. So if you want to check it out, just head over there. I will assemble the mouse on the robot this weekend. I will post the updates here in the news section.
I published the article for the issue in the pose calculation and it has a brief descriptiton of the issue. Link to the article
I successfully migrated all my projects pages on wordpress, you can see them.
Transferring my projects to WordPress. Site migration ETA: today-saturday
I installed wordpress on my site, so I won’t have to write the plain html by hand. Not all pages are available currently. Temporary outages may happen.