Machine Learning, Robotics & Control

Bing Ai

https://sites.google.com/a/utexas.edu/bing-ai/industrial-experience

Research Scientist - Artificial Intelligence & Autonomous Vehicles Ford Motor Company, E-mail: ****.**@******.***

EDUCATION

M.S. Computer Science The University of Texas at Austin, Austin, Jun. 2014 to Apr. 2017 Ph.D. Mechanical Engineer University of Connecticut, Jan. 2013 to May 2014 (incomplete) Ph.D. Controls Huazhong University of Science and Technology Sept. 2008 to Jun. 2012 RESEARCH INTERESTS

Computer Vision, GANs, Reinforcement Learning, Autonomous Vehicles, Robotics, Controls TECHNICAL SKILLS

C/C++, Python, Matlab, Java, Haskell, LabVIEW, X86 Assembly Language Programming INDUSTRIAL EXPERIENCES

Ford Motor Company, Dearborn, MI Apr. 2017 to now

I am honored to be one of very few selected members in the Henry Ford Technical Fellow’s Artificial Intelligence group. This group provides technical support directly for Ford Senior Executive Team on Artificial Intelligence. Perception for Autonomous Vehicle

• Implemented several Computer Vision (CV) Algorithms (including both one-stage & two-stage object detection/segmentation methodologies) for autonomous vehicle perception project.

• Single image depth estimation using CV technics. Scene Generation for Autonomous Vehicle Simulation

• Implemented several Generative Adversarial Networks (GANs) to generate new driving scenarios. Controls & Decision Making for Autonomous Vehicle

• Designed/Implemented the FIRST dual mode Model Predictive Controller (MPC) for Ford Autonomous Vehicle in Baidu’s Apollo platform.

• Designed/Implemented several Deep Reinforcement Learning (Deep RL) Algorithms for autonomous vehicle highway lane switching.

Anomalous Sounds/Noises Detection

• Designed deep Neural Network architectures to detect anomalous sounds for Autonomous Vehicle. Autonomous Vehicle Platform Project

Launched: P/T Strategy Development - Maximum Torque Estimation: Goal: to provide estimates of maximum wheel torque that accurately predict what the powertrain can deliver at current vehicle speed and at specified vehicle speeds. This speed-torque information will be used by Virtual Driver System to make its planning decision.

• Maximum Wheel Torque Estimation at current vehicle speed:

– Reduced estimation error from more than 10% (strategy in production) to within 5%.

• Maximum Wheel Torque Estimation at specified vehicle speeds:

– Developed/Implemented the new strategy from scratch. Robot Training Academy Inc., MD May 2016 to Aug. 2016 G1 Project:

Demos: Tea Time; Rainy Night; Get Food from Refrigerator; Heat Food in Microwave; Clean Table

• One of nine founding members of Robot Training Academy Inc.

• The only member in the control team

– Designed and implemented various grasp motion procedures for both the parallel grasp robotic arm and the ReFlex hand, allowing them to fluidly grasp more complex objects.

– Wrote software to identify all possible grasp points for objects and developed heuristics for choosing the optimal grasp strategy based on the objects’ attributes.

– Designed and implemented control algorithms for “Ridgeback,” Baxter’s autonomous omnidirectional movable platform.

ACADEMIC EXPERIENCES

I made fundamental contributions to improve the exponentially weighted moving average (EWMA) algorithm. I contributed significantly to time-delayed control systems & run-to-run control. I also greatly helped to pioneer the introduction of stochastic control theory as a means of analyzing the semiconductor manufacturing processes. I have published 14 peer reviewed scholar works with varied experiences in HVAC systems, battery charge/discharge optimal control, robotic and visual servoing.

Time-Delayed Robotic Systems Aug. 2014 to Jan. 2016

• I have made fundamental contributions to time-delayed LTI systems where I mathematically proved and experimentally verified that time delay can be beneficial for some systems.

– Developed an analytical method to obtain the stability region for time-delayed robotics systems.

– Proposed a counter-intuitive methodology to improve the system stability and performance. Visual Servoing Control May 2014 to Aug. 2014

• Designed visual servoing control algorithms for the “Dreamer” robot. Cyber-Physical Infrastructure for the Smart City Jan. 2013 to Nov. 2013

• Deployed sensor networks in Electromechanical Systems lab and Engineering II building.

• Proposed an ARHMM algorithm to detect spatial and temporal distribution of building occupancy. Timed-Delayed Semiconductor Manufacturing Process Modeling and Control Jul. 2009 to Aug. 2009

• Pioneered the introduction of stochastic control theory to analyze the semiconductor manufacturing processes.

• Established mathematic models for Time-Delayed Semiconductor Manufacturing processes.

• Developed stochastic stability theory for a system with random delay. Semiconductor Manufacturing Process Control Mar. 2008 to Dec. 2011

• Developed mathematic models for mixed products run-to-run semiconductor manufacturing processes.

• Proposed different algorithms to improve the fault tolerance of the mixed products manufacturing processes.

• Developed different control algorithms to optimize the semiconductor manufacturing processes.

Contact this candidate