Upgrade T265 firmware to 0.2.0.926

[radfordi]

• Fix map load hangs
• Fix USB serial number (remove trailing zeros)
• Support for remove_static_node

[dorian3d]

Working fine for me on Mac, including the new API

[neilyoung]

I wasn't aware that there was a "map load hang", never seen this. But I'm still very unlucky with this #5683. This seems to be not fixed, right?

[neilyoung]

@radfordi Could you please elaborate on remove_static_node Having applied #5687, can't find it...

[dorian3d]

Hi @neilyoung. It's available in the rs2::pose_sensor interface, as set_static_node and get_static_node. If you have a rs2::pipeline object to access a T265 camera, you may use it like this:

rs2::pipeline pipe;
... // start pipeline
auto tm_sensor = pipe.get_active_profile().get_device().first<rs2::pose_sensor>();
tm_sensor.set_static_node("node", ...);
tm_sensor.remove_static_node("node");

[neilyoung]

@dorian3d Thanks. Strange. Something must have gone wrong with my merge of the PR. Need to double check.

[dorian3d]

The files that are updated to implement that API are in this commit.

[neilyoung]

@dorian3d Arghh... Too many devices... I did apply the patch on my Raspberry Pi via SSH, but was searching for the changes on my Mac :))

[neilyoung]

Got this on my Mac while making:

~/Documents/librealsense/build $ make
-- Internet connection identified
-- Info: REALSENSE_VERSION_STRING=2.32.1
-- Setting Unix configurations
-- using RS2_USE_LIBUVC_BACKEND
-- pybind11 v2.2.1
-- Could NOT find apriltag (missing: APRILTAG_INC APRILTAG_LIB) 
-- Unable to find apriltag library, skipping pose-apriltag example
-- Building with TM2
-- Fetching recommended firmwares:
-- D4XX_RC_VERSION: 5.12.1.0
-- D4XX_FW_VERSION: 5.12.1.0
-- SR3XX_FW_VERSION: 3.26.1.0
-- T26X_FW_VERSION: 0.2.0.926
-- http://realsense-hw-public.s3-eu-west-1.amazonaws.com/Releases/RS4xx/FW/D4XX_FW_Image-5.12.1.0.bin
-- Download firmware 0;"returning early; file already exists with expected SHA1 hash" for D4XX_FW_Image-5.12.1.0.bin
-- http://realsense-hw-public.s3-eu-west-1.amazonaws.com/Releases/RS4xx/FW/D4XX_RC_Image-5.12.1.0.bin
-- Download firmware 0;"returning early; file already exists with expected SHA1 hash" for D4XX_RC_Image-5.12.1.0.bin
-- http://realsense-hw-public.s3-eu-west-1.amazonaws.com/Releases/SR300/FW/SR3XX_FW_Image-3.26.1.0.bin
-- Download firmware 0;"returning early; file already exists with expected SHA1 hash" for SR3XX_FW_Image-3.26.1.0.bin
-- http://realsense-hw-public.s3.amazonaws.com/Releases/TM2/FW/target/0.2.0.926/target-0.2.0.926.mvcmd
-- Download firmware 0;"No error" for target-0.2.0.926.mvcmd
-- Configuring done
CMake Error: cannot write to file "/Users/decades/Documents/librealsense/build/src/gl/CMakeFiles/Export/lib/cmake/realsense2-gl/realsense2-glTargets.cmake": Permission denied
CMake Warning (dev):
  Policy CMP0042 is not set: MACOSX_RPATH is enabled by default.  Run "cmake
  --help-policy CMP0042" for policy details.  Use the cmake_policy command to
  set the policy and suppress this warning.

  MACOSX_RPATH is not specified for the following targets:

   realsense2-gl

This warning is for project developers.  Use -Wno-dev to suppress it.

-- Generating done
CMake Generate step failed.  Build files cannot be regenerated correctly.
make: *** [cmake_check_build_system] Error 1

Resolved by removing the build folder and starting from scratch

[radfordi]

CMake Error: cannot write to file "/Users/decades/Documents/librealsense/build/src/gl/CMakeFiles/Export/lib/cmake/realsense2-gl/realsense2-glTargets.cmake": Permission denied

"Permission denied" is a local issue. Maybe you compiled once as root, then as a user or vv?

[neilyoung]

Possibly yes. I recall to have once issued "sudo make install", expecting to see the install only, but got a partial make (including cmake). Forgot that. Thanks for reminding me. No issue.

[neilyoung]

@radfordi Hi Jim, while playing around with the new image I think I found a little quirk: It is NOT possible to set_static_node, if the confidence is less than 3. That's OK. But on the other hand: If you load a map with a static node contained and the algorithm re-projects, I have seen re-projection notifications and a successful get_static_node while the confidence was 2. Isn't that inconsistent a bit? Shouldn't a re-projection only happen with at least confidence 3 and get_static_node only return OK, if the confidence is 3 too?

[radfordi]

@neilyoung, requiring a tracking confidence of 3 to create a static node was an explicit decision on our part and is not a recent change. It was implemented to avoid creating static nodes in places where there aren't enough features for us to track or where we don't have a consistent enough local map. We allow get static node when the confidence is low as it doesn't have a cumulative future effect in the same way that set static node does.

[neilyoung]

requiring a tracking confidence of 3 to create a static node was an explicit decision on our part and is not a recent change.

I know.

Thanks for clarification.

[ev-mp]

To be published in v2.32.1