[Dprglist] RealSense T265

[Patrick R. Michaud]

FWIW, my team 7172 has also been playing with this camera, and in our indoor (living room) environment we found that pointing the camera upward worked _really_ well.  There were ceiling fans and ceiling lights that it could detect; when we did that we were getting phenomenally good readings (like to within 1cm after various moves and driving around).

We haven't been as successful with the T265 in an outdoor environment -- probably too much ambient light for the sensors.

The gyro on the T265 seems to be very good; in our tests it's drifted a lot less than the BNO055 sensors that are in our control hubs (over a two minute interval).  I don't know to what degree it's incorporating data from the IMU sensors, but it seemed to work out really well for us.

Pm


On Sat, Apr 10, 2021 at 08:18:54AM +1200, Murray Altheim via DPRGlist wrote:
> Howdy,
> 
> The fact that it's a camera with an IMU and has been designed to incorporate
> odometric data, is programmable in Python or C/C++, and works on Ubuntu
> (which I'm assuming means is compatible with other Debian distributions of
> Linux like Raspberry Pi OS), would suggest that supplying this camera with
> the data from one's motor encoders and some kind of obstacle avoidance (IR/
> bumpers), one would have most of a robot's sensors.
> 
> And sells for US$200.
> 
> Is that basically true? If so, that's pretty compelling.
> 
> In theory the robot might not need the obstacle avoidance, but that would
> seem to be putting way too much onto the camera (i.e., now it needs to be
> able to recognise rocking chairs, shoes, banana peels, etc.).
> 
> As for drifting in modern houses, I've been drifting in a very non-modern
> house for about ten years, so the robot would feel right at home. Believe
> me, we have plenty of junk everywhere for the robot to look at. It won't
> get lost. But try to find the car keys or the hot water bottle and it
> will fail, just like we do.
> 
> [I'm almost a bit sad that I just yesterday put in a very expensive order
> from ServoCity (US$120 for shipping!) for 1/6th of the parts for a triple-
> bogie Mars rover -- as an experiment. It'd probably have been a much safer
> investment buying this camera...]
> 
> Cheers,
> 
> Murray
> 
> On 10/04/21 7:52 am, Karim Virani via DPRGlist wrote:
> > It seems to work pretty well in a complex environment. the team decided
> > not to use it because it drifts in environments where there are mostly
> > large planar surfaces - just not enough detail. examples include modern
> > homes with spare furnishings, gymnasiums. so i can't say we have extensive
> > experience with it. we haven't tried it outside either, though we've
> > played with the companion d345 which works well as an outdoor obstacle
> > avoidance sensor.
> > 
> > the main example they advertise is basically a turtlebot with both
> > cameras working together. we haven't tried that configuration.
> > 
> > On Fri, Apr 9, 2021 at 12:00 PM David Anderson via DPRGlist
> > <dprglist at lists.dprg.org <mailto:dprglist at lists.dprg.org>> wrote:
> > 
> >   Depth-sense Camera + IMU with built-in pose and SLAM.
> >   https://www.intelrealsense.com/tracking-camera-t265/
> >   https://dev.intelrealsense.com/docs/code-sample
> > 
> >   Pretty cool way to do odometery-like measurements if it works. Anybody
> >   have any opinions and/or experience with this device?
> > 
> >   thanks,
> >   dpa
> ...........................................................................
> Murray Altheim <murray18 at altheim dot com>                       = =  ===
> http://www.altheim.com/murray/                                     ===  ===
>                                                                    = =  ===
>     In the evening
>     The rice leaves in the garden
>     Rustle in the autumn wind
>     That blows through my reed hut.
>            -- Minamoto no Tsunenobu
> 
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at lists.dprg.org
> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org