Got an email ad for this:
Uses DTOF to be less susceptible to color/material reflectivity.
Seems pretty nice, but not cheap at $399.
/K
2 Likes
Does that mean the transmitting diode becomes the detector?
When our RPILIDAR beam hits my black UPS, there are three to five returns, centered around the normal to the bot. For 720 data measurements, that means maybe one or at most 2 degrees off the norm is detected. If the incidence angle due to the transmitting diode not being co-located with the detector is perhaps 1 degree, the loss of return is maybe two returns on one side of the norm and about 1 more inch of surface at 5 feet. That is not enough better to perform angle of the wall calculation, and only two more data points to do localization - I think the bigger improvement comes from 4x more data points, BUT the RPi might already be struggling with 720 ranges 9-10 times a second, it seems having four times that number data points to process (and send over the net as a msg) is a possible issue.
Having 3000 data points instead of 720 around the bot should improve the SLAM alot if the RPi can keep up.
1 Like
Yeah - I wonder about that. Given the price I’m not likely to check it out. That said, it’s pretty remarkable tech at that price point.
/K
2 Likes
The tech I think is most remarkable is the stereo depth camera - what folks are doing with that $80 camera - localization and 3D mapping, not just 2-D plane for walls and obstacles, full 3D mapping in the forward quadrant. The Oak-D-Lite seems like it will be able to give Dave “super powers”. (If I can figure out how to use those super capabilities.)
2 Likes
That is amazing tech. The one thing LIDAR has going for it is a 360 degree view. I guess you could have 4 cameras, but then your cost is about the same. And I don’t know if I’ve seen that. I also wonder how the depth cameras do with mirrors (of course LIDAR has its own challenges with mirrors). That’s where redundancy with other systems (e.g. ultrasound) really comes in.
/K
2 Likes