test dataset

[chengwei0427]

Hi, @HViktorTsoi

Great work! Are there any publicly available datasets for SLAM testing?

[HViktorTsoi]

Yes！We are uploading the datasets in the next few days

[chengwei0427]

Thanks for your work, and looking forward to the dataset being open-sourced.

[chengwei0427]

Hi, @HViktorTsoi @yuzehh
I downloaded the data you shared in your alpha_lidar repository. However, since the LiDAR keeps rotating from the moment the bag is opened, I have tested FastLIO2, Point-LIO, and IG-LIO, but none of them are running correctly. Could you please advise on how to make this data work? Is dynamic initialization necessary to get it running?

[HViktorTsoi]

Hi, the datasets may be challenging for other LIOs to initialize due to rapid rotation from the begining. αLiDAR has a non-static initialization strategy when starting state estimation. Please test with our uploaded code and see the initialization part.

Also please check if the initial LiDAR-IMU extrinsics are correct.

[satyajitghana]

@chengwei0427 i think you'll also need to fuse encoder data as well to state estimation in other slam algorithms

alpha_lidar/software/alpha_lidar_ws/src/state_estimation/src/voxelMapping.cpp

Lines 1196 to 1208 in 4f7c6ef

	// fuse encoder angles with odometry, transform odomtry to a fixed frame (w.r.t carrier)
	if (encoder_fusion_en){
	nav_msgs::Odometry odometry_encoder_body =
	uncoupled_encoder_fusion(lidar_end_time_prev, state_point_prev);
	// XXX: use previous frame to avoid the high probability of odometry fall outside of encoder queue
	// TODO: perform encoder fusion in seperated thread with fixed frequency
	state_point_prev = kf.get_x();
	lidar_end_time_prev = lidar_end_time;
	if (!(odometry_encoder_body.child_frame_id == "-1" || odometry_encoder_body.child_frame_id == "-2"))
	{
	pubOdomEncoderBody.publish(odometry_encoder_body);
	}
	}