Jan 23 – Jan 30 Week-27
Achieved: LSD SLAM implementation(Part-1)
Description: Developing an IO Wrapper for lsd_slam_core. implementation in OpenCV. The core library is an implementation of the motion estimation for the quadrocopter.
Feb 1 – Feb 6 Week-28
Achieved: LSD SLAM implementation(Part-2)
Description: Once the core library is successfully integrated, openFABMAP will be integrated to the system for detecting loop-closures. For Graph optimization problem g2o framework will be used after integrating it with OpenCV.
Feb 7 – Feb 14 Week-29
Achieved: Developing the lsd_slam_visualization submodule
Description:PCL(Point Cloud Library) will be used to represent 3D visualization for the map generated by the SLAM system. It must be noted that since a quadrocopter has limited computational resources in terms of memory and processing speed a library must be used which requires minimal amount of onboard resources.
Feb 15 – Feb 29 Week-30,31
Achieved: Unit testing LSD-SLAM
Description: LSD-SLAM will be tested on both publicly available benchmark sequences like the one developed by University of Freiburg, as well as live using the monocular camera onboard the quadrocopter. The functional aspects of the SLAM module will be documented in doxygen.
March 1 – March 7 Week-32
Achieved: Dense Visual SLAM for RGB-D camera
Description: With the freenect and openni library as an IO wrapper, I will be integrating dvo_core library with OpenCV. Once the core library is ported to OpenCV, visualization module is developed using the PCL visualizer to represent the dense 3D map.
March 8 – March 15 Week-33
Achieved: Visual odometry and Sensor model integration
Description: After implementing all the SLAM systems, sensor model from the IMU is integrated.
March 16 – March 23 Week-34
Achieved: Configuring Dense optical flow
Description: Dense optical flow will be used to implement a motion sensing module on the Raspberry pi camera. calcOpticalFlowSF() API will be used.
March 24 – March 31 Week-35
Achieved: Unit Testing for visual odometry and tracking module
Description: Respective modules will be tested with the standard benchmarks available. For the visual odometry from monocular camera, KITTI dataset would be used and for RGB-D camera dense visual odometry ICL-NUIM dataset and TUM RGB-D dataset would be used.
April 1 – April 7 Week-36
Achieved: Navigation module
Description: Major task would be to develop a state-estimation module which includes an OpenCV implementation of PTAM. Then the Linear Kalman filter class in OpenCV must be modified to Extended Kalman filter to incorporate the data parameters from the IMU sensor. Finally learned 3D feature visual Map will be made using the OpenGL library.
April 8 – April 15 Week-37
Achieved: Unit Testing for Navigation module
Description: The navigation module will be tested by letting the quadrocopter complete large variety of different figures like Rectangle, Haus vom Nikolaus. Documentation will be written on the usage of these APIs.
April 16 – April 23 Week-38
Achieved: Generating OpenCV python bindings and integration testing.
Description: Experimental quadrocopter whose specifications are mentioned in the first page will be used for testing purposes. The generated python APIs would be us in the Multiwii control program. Any bugs pertaining will be addressed and usage documentation will be committed.