Mountain View, California, United States
http://www.github.com/hanbrianlee Long Term Goal: Technical specialist -> Owner/CEO/President Specialties: Deep Learning Perception/Localization Autonomous Driving Technologies Sports
Working on various innovative deep Learning solutions utilizing various AI/ML techniques, such as CNN, Transformers, foundational models, custom loss functions, custom layers. Design new algorithms for proof of concept/research and production environment. Paper publications and patenting. Solving deep learning based online scene mapping and end-to-end mapless driving, fusing multimodal information such as cameras, lidars, lower quality maps, foundation models, other modes of information. Leading abstract localization development using deep neural networks. Supporting adversarial AI work. Solving research and production program problems with multiple algorithms including machine learning/deep learning (CNN, transformer, foundational models, autoencoders, etc), bayesian filters (particle filters, kalman filters), real-time data association using algorithms such as Hungarian method and ICP Estimating map bias using Gaussian Processes Researching AI/ML related to perception/auto-annotation - Tensorflow and PyTorch Lidar, radar, camera, and other sensors integration and intrinsic/extrinsic alignments
Led several 12+ large and small projects to completion from design stage realizing cost savings along the way, complying to industrial standards as well as handled negotiation with clients and vendors.
Along with several co-founders, I founded SPEC on September 2010 and we consist mostly of undergraduates but also include some people who work in the industry like myself. We have been working on several projects, hosting events, conducting presentations for funding, and providing seminars trying to achieve multiple objectives on SFU: - Managed and mentored 10+ engineers on 4 robotics projects. - Conducted several presentations to raise funds and successfully acquired funding from ESSS, Dean of Engineering, and other private investors. - Provided engineering training sessions to members such as different techniques on soldering, PWM control theories, electrical theories, etc. - Led design meetings and wrote functional specifications. Created gantt charts for project progress management. - UHDMS project: I reverse-engineered existing remote controlled cars to retrofit it with MSP430 and a custom PCB with full H-bridge PWM control, enabling 4 dimensional speed control, steering with servo motor, obstacle identification with proximity sensors, and wireless connectivity with SimpliciTI. These autonomous cars were able to turn corners on custom made tracks. This project's goal was to emulate the kind of future V2V and V2I interactions in the autonomous driving era. - MROMS: Multiple wheeled robots taking on various search/rescue ojbectives - Solar Inverter: I designed the PWM full-bridge inverter circuit with 12V solar inverter cells and 100W 1:10 transformer. Small appliances were run for project demonstrations at club days - Designed the electronic architecture for Titanoboa project for eatArt. I delivered schematics and PCB layouts/gerber files using Altium Designer. Incorporated MOSFETs for PWM control of pneumatic valves for slither control and SN65HVD233 for CAN transceivers.
As part of an exciting start-up engineering team, designed V2G-enabled 100% Battery electric vehicle Put on multiple disciplinary hats and performed various hands-on work - Designed and programmed core software that controls the vehicles, sensor processing, V2I controls, motor controls, HMI displays. - Designed the vehicle control system on ECU from scratch utilizing rugged automotive rapid prototyping platform (Mototron) and real-time operating system/model based design package (motohawk). Design work was done in MATLAB/Simulink/Stateflow - Calibrated various PID parameters and gains using Mototune - I assisted in custom PCB designs, schematic capture, and assembly (reflow soldering and manual soldering). Programmed embedded C program in freescale codewarrior environment. Worked with freescale MC9S08DV60 micro-controllers, Vacuum Florescent Display Module(M0116SD), serial in-out registers(74ls164) for resistor array for fooling existing Ford fuel gauge, etc. Serial interfaces done in I2C, SPI. - Applied debouncing for buttons, hysteresis for pressure sensors / temperature sensors - Harnessed electrical wires, installed, and ECU-integrated LiFePO4 batteries, battery management system (Valence), Coulomb J1772 charging station, Delphi 2.2kW DC/DC converter, EDN 6.6kW charger, 100kW UQM motor controller, Water coolant pump, vacuum pump, electric heater, 6kW grid-tie inverter, etc. - Tested pre-charge controller using oscilloscope - Picked out Bussman fuses and Gigavac magnetic contactors - Established in-vehicle Modbus RS485 and CANBus networks. - Performed FMEA with the team to improve reliability of the system - Performed SIL, HIL, and road tests - Troubleshot various problems