fix: update trt api

[wep21]

Description

closes #1436

Pre-review checklist for the PR author

The PR author must check the checkboxes below when creating the PR.

• [x] I've confirmed the contribution guidelines.
• [x] The PR follows the pull request guidelines.

In-review checklist for the PR reviewers

The PR reviewers must check the checkboxes below before approval.

• [ ] The PR follows the pull request guidelines.

Post-review checklist for the PR author

The PR author must check the checkboxes below before merging.

• [ ] There are no open discussions or they are tracked via tickets.

After all checkboxes are checked, anyone who has write access can merge the PR.

[codecov[bot]]

Codecov Report

Merging #1472 (0bc7e5b) into main (f8b090b) will decrease coverage by 0.02%. The diff coverage is 0.00%.

@@            Coverage Diff             @@
##             main    #1472      +/-   ##
==========================================
- Coverage   10.73%   10.71%   -0.03%     
==========================================
  Files        1115     1115              
  Lines       78831    78987     +156     
  Branches    18561    18561              
==========================================
  Hits         8464     8464              
- Misses      61507    61663     +156     
  Partials     8860     8860              

Flag	Coverage Δ		*Carryforward flag
differential	0.00% <0.00%> (?)
total	10.72% <0.00%> (ø)		Carriedforward from f8b090b

*This pull request uses carry forward flags. Click here to find out more.

Impacted Files	Coverage Δ
...idar_apollo_instance_segmentation/src/detector.cpp	0.00% <0.00%> (ø)
...lidar_centerpoint/lib/network/tensorrt_wrapper.cpp	0.00% <0.00%> (ø)
perception/tensorrt_yolo/lib/src/trt_yolo.cpp	0.00% <0.00%> (ø)
...tion/traffic_light_classifier/utils/trt_common.cpp	0.00% <0.00%> (ø)
...tion/traffic_light_classifier/utils/trt_common.hpp	0.00% <ø> (ø)
...raffic_light_ssd_fine_detector/lib/src/trt_ssd.cpp	0.00% <0.00%> (ø)
... and 1 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[yukkysaito]

is it supported to work on x86 and nvidia gpu not only Jetson AGX?

[yukke42]

The packages have failed to be built in the autoware environment.

Starting >>> lidar_apollo_instance_segmentation
--- stderr: lidar_apollo_instance_segmentation                                
/home/yusuke/ghq/github.com/tier4/pilot-auto.xx1/src/autoware/universe/perception/lidar_apollo_instance_segmentation/src/detector.cpp: In constructor ‘LidarApolloInstanceSegmentation::LidarApolloInstanceSegmentation(rclcpp::Node*)’:
/home/yusuke/ghq/github.com/tier4/pilot-auto.xx1/src/autoware/universe/perception/lidar_apollo_instance_segmentation/src/detector.cpp:63:13: error: ‘class nvinfer1::IBuilderConfig’ has no member named ‘setMemoryPoolLimit’
   63 |     config->setMemoryPoolLimit(nvinfer1::MemoryPoolType::kWORKSPACE, 1 << 30);
      |             ^~~~~~~~~~~~~~~~~~
/home/yusuke/ghq/github.com/tier4/pilot-auto.xx1/src/autoware/universe/perception/lidar_apollo_instance_segmentation/src/detector.cpp:63:42: error: ‘nvinfer1::MemoryPoolType’ has not been declared
   63 |     config->setMemoryPoolLimit(nvinfer1::MemoryPoolType::kWORKSPACE, 1 << 30);
      |                                          ^~~~~~~~~~~~~~
make[2]: *** [CMakeFiles/lidar_apollo_instance_segmentation.dir/build.make:76: CMakeFiles/lidar_apollo_instance_segmentation.dir/src/detector.cpp.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:82: CMakeFiles/lidar_apollo_instance_segmentation.dir/all] Error 2
make: *** [Makefile:141: all] Error 2
---
Failed   <<< lidar_apollo_instance_segmentation [10.0s, exited with code 2]

Summary: 10 packages finished [13.7s]
  1 package failed: lidar_apollo_instance_segmentation
  1 package had stderr output: lidar_apollo_instance_segmentation

❯ sudo dpkg -l | grep libnvinfer
hi  libnvinfer-dev                                      8.2.4-1+cuda11.4                     amd64        TensorRT development libraries and headers
hi  libnvinfer-plugin-dev                               8.2.4-1+cuda11.4                     amd64        TensorRT plugin libraries
hi  libnvinfer-plugin8                                  8.2.4-1+cuda11.4                     amd64        TensorRT plugin libraries
hi  libnvinfer8                                         8.2.4-1+cuda11.4                     amd64        TensorRT runtime libraries

[wep21]

This PR requires https://github.com/autowarefoundation/autoware/pull/2766.

[yukke42]

@wep21 I checked the lidar_centerpoint package runs successfully with https://github.com/autowarefoundation/autoware/pull/2766. But there are no outputs from the lidar_apollo_instance_segmentation package by the error [lidar_apollo_instance_segmentation_node-29] ERROR: 3: [executionContext.cpp::executeV2::521] Error Code 3: API Usage Error (Parameter check failed at: runtime/api/executionContext.cpp::executeV2::521, condition: !mEngine.hasImplicitBatchDimension(). So traffic_light_classifier may also have the same error. Cloud you please check this error?

[wep21]

@yukke42 Could you try https://github.com/autowarefoundation/autoware.universe/pull/1436 first? It may be better to keep backward compatibility for multiple l4t version e.g. 32.6(tensorrt 8.2) and 34.1(tensorrt 8.4).

[wep21]

@NilaySener I added modifications to compile without adding -Wno-deprecated-declarations except for lidar_apollo_instance_segmentation. Could you try and review this PR?

[wep21]

@yukke42 I added more modifications to build with tensorrt 8.2. Could you try to run with a rosbag again? I reverted changes for lidar_apollo_instance_segmentation because explicit batch is not supported for caffe parser. https://github.com/NVIDIA/TensorRT/issues/470

[NilaySener]

@NilaySener I added modifications to compile without adding -Wno-deprecated-declarations except for lidar_apollo_instance_segmentation. Could you try and review this PR?

@wep21 I tried to compile this PR and it didn't give a deprecation error on Jetson AGX-Orin.

[wep21]

@NilaySener Thank you for confirming it. Your PR is closed due to merging this PR, but thank you for your contribution.