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

About

Research ideas/rapid prototyping of slam systems

Languages

Language:C++ 70.4%Language:Cuda 11.8%Language:GLSL 10.0%Language:Python 4.3%Language:CMake 1.9%Language:Dockerfile 0.8%Language:Shell 0.6%Language:C 0.1%