This Project Consists of Creating a point cloud map using Lego-Loam SLAM algorithm on NUANCE car for part of  northeastern campus. For that we modified existing Lego-Loam to work with our dataset.

Lightweight and ground optimized lidar odometry and mapping system for ROS compatible UGVs (used on NUance car for our project). The system takes point cloud data from a VLP-16 Lidar and outputs 6D pose estimation in real-time. The proposed system seeks to improve efficiency and accuracy for ground vehicles while vastly reducing computation time to preceding LOAM applications and for practical robotic applications.

For this project we collected data into a Rosbag through sensors on NUANCE car around Boston campus which include LIDAR or specifically Velodyne’s Puck lidar sensor(VLP 16), VectorNAV 9 deg IMU,GPS and Thermal camera.

Image of NUANCE car is as below:

We tried to create a map of campus with the data collected using Lego-loam and also tried to do loop closure to check the consistency of the map.

below you can see the video of lego-loam implementation :

The above video shows a fully detailed video of Lego loam implementation, below images in short tells the output images of the code,we also did Aloam and SC-LegoLoam for comparison and the detailed ppt is attached Here.

below image you can see the trajectory change with loop closure on and off :

With this project, we concluded that Lego-Loam can handle loop closures over small distances and in a static environment, Lego-Loam falis in the dynamic environment and long distances.

all the modified code is uploaded on GitHub.