PETER ONDRUSKA

About

I am a researcher and entrepreneur working at the intersection of machine learning, robotics, and computer vision. I am currently the Head of Research at Lyft, Level 5 division, where I lead a team working on leapfrog self-driving cars technology.

Since age 15, I had been an active competitive programmer representing the country at international math and computer science competitions. I did my Ph.D. at Oxford Robotics Institute and worked at Google, Facebook, Microsoft Research, and others.

In 2016 I founded and served as CEO of Blue Vision Labs, where we raised over $17M and developed the first city-scale SLAM system for collaborative Augmented Reality and Robotics. In 2018 we launched our SDK product and shortly after got acquired by Lyft to use our technology to build self-driving cars.

I specialize in building teams and deep technology. I like to attend research conferences, organise ML competitions, blog and do angel investing. If you are interested in my work or are building something great that I can help with, do get in touch.

Rising stars in self-driving

Rising stars in self-driving

Forbes 30 under 30

Forbes 30 under 30

Blue Vision Labs acquired by Lyft

Blue Vision Labs acquired by Lyft

Blue Vision Labs raises $14M from GV, Accel and Horizon Ventures

Blue Vision Labs raises $14M from GV, Accel and Horizon Ventures

Blue Vision Labs selected among Tech Pioneers

Blue Vision Labs selected among Tech Pioneers

Blue Vision Labs launches ARCloud

Blue Vision Labs launches ARCloud

100 most influential people shaping UK tech

100 most influential people shaping UK tech

Blue Vision Labs cover story

Blue Vision Labs cover story

Recent projects

  • Level 5 ML planning stack

    The Autonomy 2.0 autonomy stack at Lyft Level 5 / Toyota Woven Planet deployed in production in San Francisco. Instead of hand-engineering the self-driving car decisions these are learned directly from human driving demonstrations.

    Read

  • Blue Vision Labs AR Cloud

    A large-scale localisation and mapping system for outdoor augmented reality allowing shared and persistent AR experiences on consumer mobile phones. The system was deployed at scale in producing in San Francisco and London.

    Read

  • self-driving-cars.org

    A tutorial on Autonomy 2.0 paradigm based on 1. Large-scale data collection 2. data-driven simulation and 3. end-to-end differnetiable and trainable stack. Presented at CVPR 2021, accompanied by L5Kit, Level 5 open self-driving datasets and several research publications.

    Learn

  • L5Kit.org

    An open-source toolkit for building Autonomy 2.0 systems. Allows building ML-based prediction, planning and simulation from collected datasets.

    Try

  • Open self-driving datasets

    Open-source perception and prediction datasets for training self-driving ML systems. Used in Level 5 production systems, several academic publications and ML competitions.

    Download

  • Kaggle ML competitions

    ML perception and prediction competition organised in collaboration with Kaggle. Attracting in total more than 1400 participating teams and giving $55k in prizes.

    Compete

Publications

I publish some of my work at peer-reviewed research conferences. If you want see the full list, please visit my Google Scholar page. I also help to organise various Kaggle ML competitions and workshops, tutorials at CVPR, ICRA, ICCV and NeurIPS.

 

SafetyNet: Safe planning for real-world self-driving vehicles using machine-learned policies

M Vitelli, Y Chang, Y Ye, M Wołczyk, B Osiński, M Niendorf, H Grimmett, Q Huang, A Jain, P Ondruska
(Under review)

DriverGym: Democratising Reinforcement Learning for Autonomous Driving

P Kothari, C Perone, L Bergamini, A Alahi, P Ondruska
NeurIPS 2021, Autonomous Driving Workshop

Urban Driver: Learning to Drive from Real-world Demonstrations Using Policy Gradients

O Scheel, L Bergamini, M Wołczyk, B Osiński, P Ondruska
CoRL 2021

Autonomy 2.0: Why is self-driving always 5 years away?

A Jain, L Del Pero, H Grimmett, P Ondruska

SimNet: Learning Reactive Self-driving Simulations from Real-world Observations

L Bergamini, Y Ye, O Scheel, L Chen, C Hu, L Del Pero, B Osinski, P Ondruska
ICRA 2021

What data do we need for training an AV motion planner?

L Chen, L Platinsky, S Speichert, B Osinski, O Scheel, Y Ye, H Grimmett, P Ondruska
ICRA 2021

One Thousand and One Hours: Self-driving Motion Prediction Dataset

J Houston, G Zuidhof, L Bergamini, Y Ye, L Chen, A Jain, S Omari, V Iglovikov, P Ondruska
CoRL 2020

Collaborative Augmented Reality on Smartphones via Life-long City-scale Maps

L Platinsky, M Szabados, F Hlasek, R Hemsley, L Del Pero, A Pancik, B Baum, H Grimmett, P Ondruska
ISMAR 2020

Level 5 Perception Dataset

R Kesten, M Usman, J Houston, T Pandya, K Nadhamuni, A Ferreira, M Yuan, B Low, A Jain, P Ondruska, S Omari, S Shah, A Kulkarni, A Kazakova, C Tao, L Platinsky, W Jiang, V Shet
NeurIPS 2019, Tutorial on Autonomous Driving

Visual vehicle tracking through noise and occlusions using crowd-sourced maps

MS Suraj, H Grimmett, L Platinský, P Ondrúška
IROS 2018

VALUE: Large Scale Voting-based Automatic Labelling for Urban Environments

G Dabisias, E Ruffaldi, H Grimmett, P Ondruska
ICRA 2018

Deep tracking in the wild: End-to-end tracking using recurrent neural networks

J Dequaire, P Ondrúška, D Rao, D Wang, I Posner

Ask me anything: Dynamic memory networks for natural language processing

A Kumar, O Irsoy, P Ondruska, M Iyyer, J Bradbury, I Gulrajani, V Zhong, R Paulus, R Socher
ICML 2016

End-to-end tracking and semantic segmentation using recurrent neural networks

P Ondruska, J Dequaire, DZ Wang, I Posner
RSS 2016, Workshop

Neural Robotics - A New Perspective

P Ondruska
AAAI 2016, Robotics Fellowship

Maximum entropy deep inverse reinforcement learning

M Wulfmeier, P Ondruska, I Posner

Deep inverse reinforcement learning

M Wulfmeier, P Ondruska, I Posner
CoRR 2015

Scheduled perception for energy-efficient path following

P Ondrúška, C Gurău, L Marchegiani, CH Tong, I Posner
ICRA 2015

Blog