About
Xueyang Kang

Robotic Software Engineer at Qualcomm

Get involved in the commercialization of Qualcomm new VSLAM 3.0 product based on Snapdragon QC845(Cortex-A75). Code programming and porting for EKF based fusion framework to provide prior pose for visual SLAM.

• Fused semantic label into SLAM association framework, to speed up feature search and enhance matching performance, please refer to my first paper in publication list.
• Programmed the EKF based framework, VO being loosely coupled with imu, wheel encoder and depth to improve the system robustness and accuracy.
• Ported the sensor fusion code from "ROS" to Windows for debugging and simulation purpose, Optimized latency and memory management via SSE (Streaming SIMD Extensions).

Student as Udacian

a.Involved in Robotic Software Engineer Nanodegree Program.

Term1

1. Search and Sample Resturn
2. Pick and Place
3. 3D Perception
4. Quadrotor Follows Me

Term2

1. Robotic Inference
2. Where Am I?
3. Map My World Robot
4. Reinforcement Learning based Manipulator

Certificate granted for successful graduation.

b.Involved in Flying Car Nanodegree Program.

1. Backyard Flyer
2. 3D Motion Planning
3. Nonlinear Cascaded Controller
4. EKF for Quadrotor:

Certificate granted for successful graduation.

MASTER Student at Technical University Munich

3D reconstruction via static and dynamic way, via affordable 2D Lidar, Part1, Part2.

• Setup work for hardware, and development environment on Raspberry Pi, Ubuntu PC, driver install and data acquisition.
• Manufactured a kit for 3D scanning to contain the Lidar and motor, performed 3D scan at static location.
• Programmed to synchronize the different data streams and completed calibration for the whole sensor framework.
• Fused attitude estimation from IMU with position estimation from HectorSLAM to generate 3D scan along motion.
• Desinged specific metrics implemented through Octree iterator, to evaluate the similarity and difference of two point cloud by varying means for a same scenario.

Research Assistant at Techinical University Munich

Sensor calibration and hardware setup. Due to Intellectual Property, partial code not public.

• Mitigated the stereo camera Bumblebee, and Decawave UWB driver in ROS framework.
• Established the simulator in Gazebo, Used URDF to model a physical rover, integrating the sensot Plugin to interface with Gazebo world.
• Built the synchronization framework for image pair and UWB.
• Initialize the ORB SLAM with the depth triangulated from stereo camera to improve the tracking performance.

Intership at Shanghai ABB

Robot Speed Controller Programming on FPGA

• Programmed state transition sequence for brushless motor on ZYNQ 7000 Xilinx(FPGA + ARM A9).
• Desinged a interfacing terminal to query the motor status, and configure the control mode.
• Programmed the speed PID controller on ARM A9, and tested the acceleration and deceleration performance of motor.

Bachelor Student at Hangzhou Dianzi University

• In charge of algorithm implementation, for tracking maximal output power from pv.
• Programmed the Maximal Power Point Tracking(MPPT) and extraction algorithm and realized on DSP.
• Cooperated with team members to prototype the power circuit, and Tested the hardware with DSP controller flashed with MPPT algorithm.
• Assembled the whole system with real photovoltaic cell array, and verified the performance under dynamic disturbance, like partial occlusion or extreme wheather condition like cloudy or raining.