07/13/2022
By Sokny Long
The Francis College of Engineering, Department of Electrical and Computer Engineering, invites you to attend a doctoral dissertation defense by Sharath Patil on “Differentiation Based Peak Detection for Fast and Compact LIDAR Systems.”
Ph.D. Candidate: Sharath Patil
Defense Date: Wednesday, July 27, 2022
Time: 3 to 5 p.m. EDT
Location: Southwick 240
Committee Chair (Advisor): Martin Margala, Professor, Electrical and Computer Engineering, UMass Lowell
Committee Members:
- Joyita Dutta, Associate Professor, Electrical and Computer Engineering, UMass Lowell
- Paul Robinette, Assistant Professor, Electrical and Computer Engineering, UMass Lowell
- Corey Shemelya, Assistant Professor, Electrical and Computer Engineering, UMass Lowell
Brief Abstract:
Autonomous vehicles are ubiquitous in several applications like warehousing, autonomous cars, etc. Autonomous cars are gaining popularity and acceptance in recent years. Autonomous vehicles rely on several sensors to help with their navigation to the destinations. This might include avoiding collisions with obstacles and mapping the surroundings. One of the main challenges for autonomous vehicles is to detect if there is an obstacle in its path and automatically maneuver the vehicle to avoid hitting an object. Several sensors are used for collision avoidance depending on the requirements - speed, resolution, cost, power, etc. and LiDAR is one such popular sensor. Efforts are under way to reduce the area and power consumption of LiDAR sensors which in turn reduce the sensor cost, enabling better system integration due to smaller size. In an effort to reduce the size and cost of the LiDAR sensor, this research work has come up with a novel approach of using differentiation to detect the peak location in the received LiDAR signal. A single chip solution that implements the proposed
approach can drastically reduce the footprint of the sensor signal chain by more than 30%, by using comparators for performing the differentiation operation. Although there is a reduction in performance at low SNR conditions, the trade-off could be well worth the savings in chip area and power achieved by this method. A novel prototype has been designed and built on a PCB to assess the distance measurement accuracy of the differentiation method compared to that of the standard method that captures full LiDAR waveforms using high-resolution, high-speed ADCs. In the proposed system, pulse detection mechanism has been implemented using an array of sample and hold capacitors followed by a comparator. The prototype also consists of a full waveform LiDAR using a high-speed ADC. The control circuitry and data processing are implemented on a Xilinx ZYNQ-based FPGA. The performance of a comparator-based pulse detection has been compared to that of an ADC-based pulse detection. Measurement results show that the comparator based system has similar performance compared to ADC-based systems except under low SNR conditions.
Bio
Sharath Patil has a Bachelors in Electronics and Communication Engineering from S. J. College of Engineering, Mysore, India. He has worked as an IC designer at Kawasaki Microelectronics at Bengaluru, India. After obtaining his Masters in Electrical Engineering from the University of Texas Arlington, Sharath worked as a Design Engineer at Melexis in Nashua, NH and at Sensata in Attleboro, MA. He is currently working as a Software Development Engineer at Mathworks in Natick, MA. In addition to keeping up with the latest research in chip design and technology, he is enthusiastic about applying AI to chip design workflows. He lives in Ashland, MA with his wife and two kids.
All interested students and faculty members are invited to attend the in-person defense.