Apoorv Singh

Tech Lead | Sr. Machine Learning Engineer | Self-driving cars | Robotics | Carnegie Mellon

View on GitHub

image


I am a Robotics graduate from the School of Computer Science (Robotics) at Carnegie Mellon University (Pittsburgh). I specialize in Computer Vision/ Machine Learning, focusing on Autonomous vehicles. I work at Motional (A self-driving car company) as a Tech Lead and Sr. Machine Learning Engineer. In my current role, I am responsible for developing transformer-based state-of-the-art ML/Computer Vision algorithms to build better and faster online perception systems. I, along with my team, ensure that the ML networks I develop are efficient enough to be deployed on our production cars as well. Our networks take multi-modal input and give runtime efficient accurate perception output (Bounding boxes, Segmentation masks, and much more).

Previously, in 2019, I worked at Aptiv as a perception intern. There, I developed a novel deep-learning-based auto-annotation tool for HD Maps. We patented the technology as well. Also, I have worked as a researcher for an autonomous truck collision avoidance project at Carnegie Mellon, sponsored by Daimler Trucks North America. I was responsible for the Perception, Tracking, and Sensor Fusion pipeline. Before coming to CMU, I worked at the RnD division of Maruti Suzuki.

I co-founded a startup at CMU, called Vera and won a small NSF grant from a business competition.

I have several patents and published research papers, mostly first authored. I also believe in giving back to the Machine Learning research community through research paper reviews and keynote tutorial sessions. When I am not doing Machine Learning, I read content on human psychology, financial investments, the Future of humanity, and macroeconomics.


Education

Carnegie Mellon University

Master’s in Robotics, School of Computer Science (2018-2020)

Specialized in Artificial Intelligence, Computer Vision, Computer Science, and Robotics.

Delhi Technological University

Bachelor’s in Technology (2011-2015)

Specialized in Artificial Intelligence and Robotics


Professional Experience

Motional (2020-Present), California USA

Tech Lead and Sr. Machine Learning Engineer (Perception)

Tech lead of 5 Machine Learning and Robotics engineers and interns to develop a state-of-the-art vision network that runs in real-time on-car. This network uses eight surround-view cameras and predicts agents in the 3D world. Authored a blog for this project, published on Motional’s portal

Deployed numerous Vision-first Deep Learning models on-car using TensorRT optimizations.

Managed Reading Group and ML literature surveys within a team of 40+ Perception Engineers.

Co-authored 13 patents and 9 selected research papers in computer vision and computer science domains.

Aptiv (Summer’19), Pennsylvania USA

Perception Intern (AI)

Developed segmentation algorithm for road intersections to auto-annotate H.D. Maps by fusing images and LiDAR intensity maps. This Project led to a patent, too.

The Success of this internship project led to the forming of a team of three full-time software engineers.

Daimler Trucks North America (2018-2020), Oregon USA

Research Collaborator (Perception Lead)

Developed a reliable high-speed on-coming collision prevention system on CARLA sim for country roads.

Vera (Fall’ 2019), Pennsylvania USA

Co-founder

Developed an AI assistant for lawyers, to help with paralegal tasks. Won an NSF grant through business competition at Carnegie Mellon.

Maruti Suzuki (2015-2018), India

Assistant Manager, Research and Development

Led a smart mobility project and cross-collaborated with Homologation teams and Japanese researchers.

MSSL Global RSA (Summer’13), South Africa

Summer intern

Developed a computer vision-based part detection and quality insurance tool that helped save on the downtime of a 6 degrees of freedom robotic arm at the paint shop.


Patents

MACHINE LEARNING - BASED FRAMEWORK FOR DRIVABLE SURFACE ANNOTATION

Inventors: Sergi Adipraja Widjaja, Venice Erin Baylon Liong, Zhuang Jie Chong, Apoorv Singh

US 11,367,289 B1

TRAINING MACHINE LEARNING NETWORKS FOR CONTROLLING VEHICLE OPERATION

Inventors: Apoorv Singh, Varun Kumar Reddy Bankiti

Application number: US 18/141,014

MACHINE LEARNING-BASED FRAMEWORK FOR DRIVABLE SURFACE ANNOTATION

Inventors: Sergi, Venice, Zhuang, Apoorv Singh

US 2023/0016246 A1

ENRICHING FEATURE MAPS USING MULTIPLE PLURALITIES OF WINDOWS TO GENERATE BOUNDING BOXES

Inventors: Jongwoo, Apoorv Singh, Varun

US 2024/0062520 A1

ENRICHING OBJECT QUERIES USING A BIRD’S-EYE VIEW FEATURE MAP TO GENERATE BOUNDING BOXES

Inventors: Jongwoo, Apoorv Singh, Varun

US 2024/0062520 A1

ENRICHING FEATURE MAPS USING MULTIPLE PLURALITIES OF WINDOWS TO GENERATE BOUNDING BOXES

Inventors: Jongwoo, Apoorv Singh, Varun

Inventors: Application number: PCT/US2023/072389

AGGREGATION OF DATA REPRESENTING GEOGRAPHICAL AREAS

Inventors: Apoorv Singh, Varun, Jeongil, Akankshya

US 2024-0125617 A1

ITERATIVE DEPTH ESTIMATION

Inventors: Akankshya, Apoorv Singh, Varun

Application number: US 18/163,708

MULTI-MODAL SENSOR-BASED NAVIGATION USING BOUNDING BOXES

Inventors: Apoorv Singh

Application number: PCT/US2023/018569

Vision-RADAR fusion for DETR-like 3D detections

Inventors: Apoorv Singh, Varun

US 2024-0127596 A1

SCENE-DEPENDENT OBJECT QUERY INITIALIZATION STRATEGY USING TEMPORAL CONSISTENCIES

Inventors: Apoorv Singh, Varun

US 2024-0127597 A1

Augment Bird’s Eye multi-view-Camera detections with Perspective View detections

Inventors: Apoorv Singh

Application number: US 63/470,125

Surround-Vision & RADAR fusion strategy using Transformers

Inventors: Apoorv Singh

Application number: US 63/505,385


Research Publications

Authors: Apoorv Singh

Conference: 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)

Authors: Apoorv Singh

Conference: 2023 IEEE Intelligent Vehicles Symposium (IV)

Authors: Apoorv Singh

Conference: Proceedings of the IEEE/CVF International Conference on Computer Vision

Authors: Apoorv Singh

Conference: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

Authors: J Park, Apoorv Singh, V Bankiti

Conference: 2023 IEEE International Conference on Image Processing (ICIP), 1930-1934

Authors: Apoorv Singh

Conference: arXiv preprint arXiv:2308.10135

Authors: Apoorv Singh

Conference: arXiv preprint arXiv:2311.18636

Authors: Apoorv Singh

Conference: Proceedings of the IEEE/CVF International Conference on Computer Vision

Authors: Apoorv Singh, G Raut, A Choudhary

Conference: Accepted in ECCV’24 (Conference yet to happen)

Authors: G Raut, Apoorv Singh

Conference: arXiv preprint arXiv:2402.16369

Authors: Apoorv Singh, A Choudhary

Conference: 2023 IEEE Conference on Artificial Intelligence (CAI), 40-41

Authors: Apoorv Singh

Conference: 2023 IEEE Conference on Artificial Intelligence (CAI), 53-54


AI Community Contributions

Publication Peer-Review

Location: NeurIPS’23, CVPR’23, ICRA’22, ICML’22, ICCV’23, IEEE CAI’23, AAAI’23 and lots more, totalling ~85.

Master’s Admission Committee

Location: Robotics Institute, Carnegie Mellon University

Years: 2023, 2024, 2025.

Technical Judge

Location: Pittsburgh Regional Science & Engineering Fair, 2023

Technical Judge

Location: FIRST Robotics Competition, 2023

Conference Program Chairs

Location: NeurIPS 2022, CVPR’23 Precognition Workshop, AAAI-2023


Keynote/ Panel Sessions

Location: IV 2023, Alaska, USA

Location: IV 2024, Jeju Islands, Korea

Location: European Conference on Computer Vision 2024 in Milan, Italy

Location: 2023 IEEE Conference on Artificial Intelligence (IEEE CAI)

Organized a workshop on autonomous vehicles with a panel discussion and two guest lecturers in Santa Clara, California, to promote research on Autonomous driving.

Co-panelists: Shivam Gautam, Fang-Chieh Chou, Aleksandr Petiushko, Sachithra Hemachandra

Location: 2023 IEEE Conference on Artificial Intelligence (IEEE CAI)