Cannot run perception_lowcost_vis.sh.

[MistreanuIonutCosmin]

Hello, and thanks for all the great work done on this framework.

I am trying to implement perception_lowcost_vis.sh.

I took the content from perception.sh as an example and ended up with:

cyber_launch start /apollo/modules/transform/launch/static_transform.launch & cyber_launch start /apollo/modules/drivers/tools/image_decompress/launch/image_decompress.launch & cyber_launch start /apollo/modules/perception/production/launch/perception_lowcost.launch &

The content of perception_lowcost.launch is the following:

cyber modules list config 1.0.0 perception_camera /apollo/modules/perception/production/dag/dag_streaming_perception_camera.dag perception 1.0.0 perception_lane /apollo/modules/perception/production/dag/dag_streaming_perception_lane.dag perception_lane 1.0.0 perception_traffic_light /apollo/modules/perception/production/dag/dag_streaming_perception_trafficlights.dag perception_trafficlights 1.0.0 motion_service /apollo/modules/perception/production/dag/dag_motion_service.dag motion_service 1.0.0

When running, I can see the lane detection working but I also got an error: [perception_lane] E0715 15:47:34.780052 12105 debug_info.cc:299] [mainboard] Failed to query camera to ground height and pitch.

I tried debugging it with gdb by running: $ gdb mainboard (gdb) run -d /apollo/modules/perception/production/dag/dag_streaming_perception_camera.dag -p perception -s CYBER_DEFAULT

I ended up with the following error which I don't know how to solve:

mainboard: cudnnBuilder2.cpp:1006: nvinfer1::cudnn::Engine* nvinfer1::builder::buildEngine(nvinfer1::CudaEngineBuildConfig&, const nvinfer1::cudnn::HardwareContext&, const nvinfer1::Network&): Assertion `it != tensorScales.end()' failed.

I just started to learn how this framework. Do you have any idea on what I am doing wrong?

Thanks!

[MistreanuIonutCosmin]

Solved the problem by using:

perception_camera /apollo/modules/perception/production/dag/dag_streaming_perception_camera.dag perception 1.0.0

instead of dag_streaming_perception_lane (I found out that the functionality is included in dag_streaming_perception_camera alongside obstacle detection).

I also change in config.pt from "modules/perception/production/data/perception/camera/params" model_type: "CaffeNet". For some reason, the default quantized int8 networks using tensorRT triggers the error above.

[daohu527]

Closed due to inactivity. If the problem persists, pls feel free to reopen it or create a new one and refer to it.