slam

SLAM system

Build

Docker

Build

docker build  --build-arg user_id=99 -t slam:dev .

Run

./run_docker.sh docker_image local_slam_repo_path local_dataset_path wlinterace(Run ifconfig and copy the wl interace id. (eg:wlxd0356714af3b)

eg:

./run_docker.sh slam:dev /home/christie/projects/slam/work/slam /home/christie/projects/slam/dataset/ wlxd0356714af3b
cd ~/slam
catkin_make -DAPP=apps/elastic_fusion.cpp

Local(incomplete)

Build system : cmake or catkin_make
Assuming Opencv and CUDA is installed, run build.sh.
eg:

./build.sh
catkin_make -DAPP=apps/elastic_fusion_file.cpp

Inputs

Working input state:

rgb info:
	img.type :16(CV_8UC3)
	range :0-255
depth info:
	img.type :2(CV_16UC1)
	range :0-5000

Dependencies

• OpenCV 3.2
• CUDA
• Pangolin
• libconfig

Benchmark

Benchmark is done on the iclnuim living room dataset

Metrics	ATE	RPE	PC Alignment score
Score	0.018483	0.0231662926318	0.00750245