Training Example: Five (part of Bosch Mobility) – Review the Data, Give Your Score & Compare to the Real AI Evaluation

Five, now proudly part of Bosch Mobility!Five is at the forefront of European efforts to develop automated driving systems. Building on our heritage and unique experience as a pioneer in this space, we are taking what we have learned from conducting groundbreaking public road trials of autonomous vehicles to inform a robust safety assurance platform that helps tackle one of the industry's most complex challenges.Following our acquisition by Bosch in 2022 now, as part of Bosch Mobility, we are driving innovation within the Automated Driving Alliance – a strategic partnership between Bosch and Volkswagen Group's CARIAD. We are leading the efforts to create, manage, and evaluate cloud-based simulation within a state-of-the-art standardised software platform, enabling our automotive partners to build automated driving systems that are safer, smarter, and more scalable than ever before. Five is at the forefront of European efforts to develop automated driving systems. Building on our heritage and unique experience as a pioneer in this space, we are taking what we have learned from conducting groundbreaking public road trials of autonomous vehicles to inform a robust safety assurance platform that helps tackle one of the industry's most complex challenges.

Faster development, quicker coverage, better analysisOffline Testing capability to develop, verify and validate ADAS and AD systemsFor test driven development and safety assurance through simulation and real world analysis.Faster, safer, more efficient developmentQuickly identify strengths and weaknesses of your stack through simulation. Validate safety before public deployment. Iterate quickly on system requirements and driving features, sharing progress with stakeholders.ScalabilityTest thousands of traffic scenarios and drive many milesCoverageGenerate diverse traffic scenarios and parameterise themEfficiencyReduce the need for costly real world fleetsVelocityShorten the development cycleOverviewScenario CreationCreate advanced and varied scenarios with multiple sequences. Set parameters on actions and variables to explore a huge range of values.Rules based evaluationUse the Rule based framework to analyse and evaluate your scenario simulations. Create these rules based on your requirementsTesting environment and workflowCreate large-scale tests for development, progression and regression testing. Manage and automate CI tests at scale during development and validation.Real world triage and assessment Evaluate on road drives. Identify, locate, review, tag, extract and triage salient data. Analyse with rule evaluation and automatically detect events.

Help us to build the future of safer autonomyExplore Open RolesWe’re engineers, scientists, developers, and designers. We’re problem-solvers, dreamers, pioneers, innovators, creatives and pragmatists. We’re new hands and accomplished pros.We drink tea and obsess over coffee. We bake, cook and eat biscuits. We cycle, run, dance, climb and read. We’re gamers, motorsport watchers, board gamers, footballers and foosballers.We’re engineers, scientists, developers, and designers. We’re problem-solvers, dreamers, pioneers, innovators, creatives and pragmatists. We’re new hands and accomplished pros.We come to work because we love solving difficult problems and by doing so, we believe the benefits of safe autonomous transportation should be felt by all.Explore Open Roles
[IMG: Perpetual Improvement]
Perpetual ImprovementWe strive to continually improve; as individuals, each other, our teams, our products and our company.Science EngineeredWe rigorously apply our mastery to intentionally design products that solve customer problems. Not more, not less.Impact EverydayWe believe in making a difference on a daily basis. This happens by empowering the right people in the right environment.
[IMG: Perpetual Improvement]
Our values are underpinned by the implicit belief that we are “thoughtful and considerate humans” which is evidenced in our communication and by virtue of the value we place on team success over individual achievement.1. Getting to know youAn initial chat with our TA team to get to know a bit more about you, your background, experience and interests. Also to tell you a bit more about us and the role.2. First interviewHere we start to dive in to your technical skills and experience to see if you're a fit for the role. We also give you an opportunity to talk to our engineers about any technical questions you may have.3. Technical testWe use a range of technical tasks to help us to assess candidate skills and technical depth and breadth. Hopefully you'll find the tasks fun at the same time.4. Final interviewWe finish our process with several interviews across areas like behaviours, architecture, coding, algorithms etc, that help us to round off our picture of you as a potential member of the Five team.5. Decision and feedbackHopefully we can conclude the process with a positive outcome and an offer to join Five. We always endeavour to provide honest and constructive feedback and truly value the time candidates invest with us.We care deeply about our people and have built a competitive reward and benefits package that caters to our team's needs. From mental wellbeing and physical health to helping to provide for our team's future financial security… and a lot more in between!Below are just a few of the benefits we offer in the UK.Contributory Pension PlanEnhanced Family Friendly PoliciesFree Life AssurancePrivate Medical InsuranceWe work on exciting, rewarding projects that make a real impact with a team that’s as talented as they are fun.KamarSoftware EngineerWorking with a great bunch of people on cutting edge technology that tangibly changes the way our world works is a great reason to get out of bed every day.RobSoftware EngineerInstead of creating science fiction, I'm excited to be using my skills to make a better future, in the real world.RussellProduct ManagerCambridgeExplore Open Roles

2026Published papersFoundation Models for Trajectory Planning in Autonomous Driving: A Review of Progress and Open ChallengesKemal Oksuz, Alexandru Buburuzan, Anthony Knittel, Yuhan Yao, Puneet K. DokaniaTransactions on Machine Learning Research (TMLR) 20262024Published papersMoCaE: Mixture of Calibrated Experts Significantly Improves Object DetectionKemal Oksuz, Selim Kuzucu, Tom Joy, Puneet K. DokaniaTransactions on Machine Learning Research (TMLR) 2024What Makes and Breaks Safety Fine-tuning? A Mechanistic StudySamyak Jain, Ekdeep Singh Lubana, Kemal Oksuz, Tom Joy, Philip H.S. Torr, Amartya Sanyal, Puneet K. DokaniaNeurIPS 2024 (main conference), and ICML 2024 (Mechanistic Interpretability Workshop)On Calibration of Object Detectors: Pitfalls, Evaluation and BaselinesSelim Kuzucu*, Kemal Oksuz*, Jonathan Sadeghi, Puneet K. DokaniaEuropean Conference on Computer Vision (ECCV) 2024 (Oral Presentation)Bucketed Ranking-based Losses for Efficient Training of Object DetectorsFeyza Yavuz, Baris Can Cam, Adnan Harun Dogan, Kemal Oksuz, Emre Akbas, Sinan KalkanEuropean Conference on Computer Vision (ECCV) 2024Segment, Select, Correct: A Framework for Weakly-supervised Referring SegmentationFrancisco Eiras, Kemal Oksuz, Adel Bibi, Philip H.S. Torr, Puneet K. DokaniaECCV 2024 (Instance-Level Recognition Workshop)Lift-Attend-Splat: Bird’s-eye-view camera-lidar fusion using transformersJames Gunn*, Zygmunt Lenyk*, Anuj Sharma, Andrea Donati, Alexandru Buburuzan, John Redford, Romain MuellerCVPR 2024 (Workshop on Autonomous Driving)Fine-tuning can cripple your foundation model; preserving features may be the solutionJishnu Mukhoti, Yarin Gal, Philip H.S. Torr, Puneet K. DokaniaTransactions on Machine Learning Research (TMLR) 2024 (Featured Certification)PreprintsAttacking Motion Planners Using Adversarial Perception ErrorsJonathan Sadeghi, Nicholas A. Lord, John Redford, Romain MuellerPre-print (2024)2023Published papersGraph Inductive Biases in Transformers without Message PassingLiheng Ma, Chen Lin, Derek Lim, Adriana Romero-Soriano, Puneet K. Dokania, Mark Coates, Philip H.S. Torr, Ser-Nam LimICML 2023Catastrophic overfitting can be induced with discriminative non-robust featuresGuillermo Ortiz-Jimenez, Pau de Jorge, Amartya Sanyal, Adel Bibi, Puneet K. Dokania, Pascal Frossard, Grégory Rogez, Philip H.S. TorrTransactions on Machine Learning Research 2023Towards Building Self-Aware Object Detectors via Reliable Uncertainty Quantification and CalibrationKemal Oksuz, Tom Joy, Puneet K. DokaniaConference on Computer Vision and Pattern Recognition (CVPR), June 2023DiPA: Probabilistic Multi-Modal Interactive Prediction for Autonomous DrivingAnthony Knittel, Majd Hawasly, Stefano V. Albrecht, John Redford, Subramanian RamamoorthyIEEE Robotics and Automation Letters, 2023Planning with Occluded Traffic Agents using Bi-Level Variational Occlusion ModelsFilippos Christianos, Peter Karkus, Boris Ivanovic, Stefano V. Albrecht, Marco PavoneIEEE International Conference on Robotics and Automation (ICRA), 2023Causal Explanations for Stochastic Sequential Multi-Agent Decision-MakingBalint Gyevnar, Cheng Wang, Christopher G. Lucas, Shay B. Cohen, Stefano V. AlbrechtAAMAS Workshop on Explainable and Transparent AI and Multi-Agent Systems, 2023Verifiable Goal Recognition for Autonomous Driving with OcclusionsCillian Brewitt, Massimiliano Tamborski, Cheng Wang, Stefano V. AlbrechtIEEE/RSJ International Conference on Intelligent Robots and Systems, 2023Sample-dependent Adaptive Temperature Scaling for Improved CalibrationTom Joy, Francesco Pinto, Ser-Nam Lim, Philip H. S. Torr, and Puneet K. DokaniaAAAI Conference on Artificial Intelligence, 2023Query-based Hard-Image Retrieval for Object Detection at Test TimeEdward Ayers, Jonathan Sadeghi, John Redford, Romain Mueller, Puneet K DokaniaAAAI Conference on Artificial Intelligence, 2023PreprintsComparison of Pedestrian Prediction Models from Trajectory and Appearance Data for Autonomous DrivingAnthony Knittel, Morris Antonello, John Redford and Subramanian RamamoorthyICRA 20232022Published papersVerifiable Goal Recognition for Autonomous Driving with OcclusionsCillian Brewitt, Massimiliano Tamborski, Stefano V. AlbrechtNeurIPS Workshop on Machine Learning for Autonomous Driving (ML4AD), 2022Parameter-free Online Test-time AdaptationMalik Boudiaf, Romain Mueller, Ismail Ben Ayed, Luca BertinettoConference on Computer Vision and Pattern Recognition (CVPR), June 2022Self-supervised Test-time Adaptation on Video DataFatemeh Azimi, Sebastian Palacio, Federico Raue, Jörn Hees, Luca Bertinetto, Andreas DengelWorkshop On Applications of Computer Vision (WACV) 2022An Active Learning Reliability Method for Systems with Partially Defined Performance FunctionsJonathan Sadeghi, Romain Mueller, John RedfordNeurIPS 2022 Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems (GPSMDMS) 2022Attacking deep networks with surrogate-based adversarial black-box methods is easyNicholas A. Lord, Romain Mueller, Luca BertinettoInternational Conference on Learning Representations (ICLR) 2022RegMixup: Mixup as a Regularizer Can Surprisingly Improve Accuracy and Out Distribution RobustnessFrancesco Pinto, Harry Yang, Ser-Nam Lim, Philip H.S. Torr, Puneet K. DokaniaNeurIPS, Conference on Neural Information Processing Systems, 2022, New Orleans, USAMake Some Noise: Reliable and Efficient Single-Step Adversarial TrainingPau de Jorge, Adel Bibi, Riccardo Volpi, Amartya Sanyal, Philip H. S. Torr, Grégory Rogez, Puneet K. DokaniaNeurIPS, Conference on Neural Information Processing Systems, 2022, New Orleans, USAANCER: Anisotropic Certification via Sample-wise Volume MaximizationFrancisco Eiras, Motasem Alfarra, M. Pawan Kumar, Philip H. S. Torr, Puneet K. Dokania, Bernard Ghanem, Adel BibiTransactions on Machine Learning Research, August 2022Perspectives on the System-level Design of a Safe Autonomous Driving StackMajd Hawasly, Jonathan Sadeghi, Morris Antonello, Stefano V. Albrecht, John Redford, Subramanian RamamoorthyAI Communications special issue on Multi-agent Systems Research in the UKAn Impartial Take to the CNN vs Transformer Robustness ContestFrancesco Pinto, Philip H. S. Torr, and Puneet K. DokaniaThe European Conference on Computer Vision (ECCV) 2022, Tel AvivSample-dependent Adaptive Temperature Scaling for Improved CalibrationTom Joy, Francesco Pinto, Ser-Nam Lim, Philip H. S. Torr, and Puneet K. DokaniaICML, Distribution-Free Uncertainty Quantification Workshop 2022, Baltimore, USACatastrophic overfitting is a bug but also a featureGuillermo Ortiz-Jiménez, Pau de Jorge, Amartya Sanyal, Adel Bibi, Puneet K. Dokania, Pascal Frossard, Gregory Rogéz, Philip H.S. TorrICML, New Frontiers in Adversarial Machine Learning (AdvML) 2022, Baltimore, USAFlash: Fast and Light Motion Prediction for Autonomous Driving with Bayesian Inverse Planning and Learned Motion ProfilesMorris Antonello, Mihai Dobre, Stefano V. Albrecht, John Redford, Subramanian RamamoorthyIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022A Human-Centric Method for Generating Causal Explanations in Natural Language for Autonomous Vehicle Motion PlanningBalint Gyevnar, Massimiliano Tamborski, Cheng Wang, Christopher G. Lucas, Shay B. Cohen, Stefano V. AlbrechtIJCAI Workshop on Artificial Intelligence for Autonomous Driving (AI4AD), 2022PreprintsDiPA: Probabilistic Multi-Modal Interactive Prediction for Autonomous DrivingAnthony Knittel, Majd Hawasly, Stefano V. Albrecht, John Redford, Subramanian RamamoorthyarXiv preprint arXiv:2210.06106, October 2022Beyond RMSE: Do machine-learned models of road user interaction produce human-like behavior?Aravinda Ramakrishnan Srinivasan, Yi-Shin Lin, Morris Antonello, Anthony Knittel, Mohamed Hasan, Majd Hawasly, John Redford, Subramanian Ramamoorthy, Matteo Leonetti, Jac Billington, Richard Romano, Gustav MarkkulaarXiv preprint arXiv:2206.11110, August 20222021Published papersAre Vision Transformers Always More Robust Than Convolutional Neural Networks?Francesco Pinto, Philip Torr, Puneet K DokaniaNeurIPS 2021 Workshop on Distribution Shifts: Connecting Methods and ApplicationsMix-MaxEnt: Improving Accuracy and Uncertainty Estimates of Deterministic Neural NetworksFrancesco Pinto, Harry Yang, Ser-Nam Lim, Philip Torr, Puneet K DokaniaNeurIPS 2021 Workshop on Distribution Shifts: Connecting Methods and ApplicationsA Step Towards Efficient Evaluation of Complex Perception Tasks in SimulationJonathan Sadeghi, Blaine Rogers, James Gunn, Thomas Saunders, Sina Samangooei, Puneet Kumar Dokania, John RedfordNeurIPS 2021 Workshop on Machine Learning for Autonomous DrivingOn Episodes, Prototypical Networks, and Few-shot LearningSteinar Laenen, Luca BertinettoTo appear at the Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021)Do Different Tracking Tasks Require Different Appearance Models?Zhongdao Wang, Hengshuang Zhao, Ya-Li Li, Shengjin Wang, Philip H.S. Torr, Luca BertinettoTo appear at the Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021)A Continuous Mapping For Augmentation DesignKeyu Tian, Chen Lin, Ser-Nam Lim, Wanli Ouyang, Puneet K. Dokania, Philip TorrTo appear at the Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021)Smart Pointers and Shared Memory Synchronisation for Efficient Inter-process Communication in ROS on an Autonomous VehicleCostin Iordache, Stephen M. Fendyke, Mike J. Jones, Robert A. Buckley2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)Interpretable Goal Recognition in the Presence of Occluded Factors for Autonomous VehiclesJosiah P. Hanna, Arrasy Rahman, Elliot Fosong, Francisco Eiras, Mihai Dobre, John Redford, Subramanian Ramamoorthy, Stefano V. Albrecht2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)GRIT: Fast, Interpretable, and Verifiable Goal Recognition with Learned Decision Trees for Autonomous DrivingCillian Brewitt, Balint Gyevnar, Samuel Garcin, Stefano V. Albrecht2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)PILOT: Efficient Planning by Imitation Learning and Optimisation for Safe Autonomous DrivingHenry Pulver, Francisco Eiras, Ludovico Carozza, Majd Hawasly, Stefano V. Albrecht and Subramanian Ramamoorthy2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)Interpretable Goal-based Prediction and Planning for Autonomous DrivingStefano V. Albrecht, Cillian Brewitt, John Wilhelm, Balint Gyevnar, Francisco Eiras, Mihai Dobre, Subramanian RamamoorthyIEEE International Conference on Robotics and Automation (ICRA), 2021, Xi'an ChinaRecurrently Estimating Reflective Symmetry Planes from Partial PointcloudsMihaela Cătălina Stoian, Tommaso CavallariCVPR 2021 Workshop on 3D Vision and RoboticsA Two-Stage Optimization-based Motion Planner for Safe Urban DrivingFrancisco Eiras, Majd Hawasly, Stefano V. Albrecht, Subramanian RamamoorthyIEEE Transactions on Robotics (T-RO), 2021Resolving Conflict in Decision-Making for Autonomous DrivingJack Geary, Subramanian Ramamoorthy, Henry GoukRobotics: Science and Systems (R:SS), 2021Progressive Skeletonization: Trimming more fat from a network at initializationPau de Jorge, Amartya Sanyal, Harkirat S. Behl, Gregory Rogez, Philip H. S. Torr, Puneet K. DokaniaInternational Conference on Learning Representations (ICLR) 2021No Cost Likelihood Manipulation at Test Time for Making Better Mistakes in Deep NetworksShyamgopal Karthik, Ameya Prabhu, Puneet K. Dokania, Vineet GandhiInternational Conference on Learning Representations (ICLR) 2021How benign is benign overfitting?Amartya Sanyal, Puneet K Dokania, Varun Kanade, Philip H.S. TorrInternational Conference on Learning Representations (ICLR) 2021Scalable FPGA Median Filtering via a Directional Median CascadeOscar Rahnama, Stuart Golodetz, Tommaso Cavallari, Philip H.S. Torr2021 IEEE 29th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM) (poster, non-archival)2020Published papersLower dimensional kernels for video discriminatorsE. Kahembwe, S. RamamoorthyNeural Networks Journal, Special Issue on Deep Neural Network Representation and Generative Adversarial Learning, 2020Continual Learning in Low-rank Orthogonal SubspacesArslan Chaudhry, Naeemullah Khan, Puneet K. Dokania, Philip H.S. TorrAdvances in Neural Information Processing Systems (NeurIPS), December 2020GDumb: A Simple Approach that Questions Our Progress in Continual LearningAmeya Prabhi, Philip H.S. Torr, Puneet K. DokaniaEuropean Conference on Computer Vision (ECCV) (oral), August 2020PaRoT: A Practical Framework for Robust Deep Neural Network TrainingEdward Ayers, Francisco Eiras, Majd Hawasly, Iain WhitesideNASA Formal Methods. NFM 2020. Lecture Notes in Computer Science, vol 12229, August 2020Beyond Controlled Environments: 3D Camera Re-Localization in Changing Indoor ScenesJohanna Wald, Torsten Sattler, Stuart Golodetz, Tommaso Cavallari and Federico TombariEuropean Conference on Computer Vision (ECCV), August 2020Calibrating Deep Neural Networks using Focal LossJishnu Mukhoti, Viveka Kulharia, Amartya Sanyal, Stuart Golodetz, Philip H S Torr and Puneet K DokaniaAdvances in Neural Information Processing Systems (NeurIPS), December 2020An Optimization-based Motion Planner for Safe Autonomous DrivingFrancisco Eiras, Majd Hawasly, Stefano V. Albrecht, Subramanian RamamoorthyRSS 2020 Workshop in Robust AutonomyMaking Better Mistakes: Leveraging Class Hierarchies with Deep NetworksLuca Bertinetto*, Romain Mueller*, Konstantinos Tertikas, Sina Samangooei and Nicholas A Lord*Conference on Computer Vision and Pattern Recognition (CVPR), June 2020Simplifying TugGraph using Zipping AlgorithmsStuart Golodetz, Anurag Arnab, Irina Voiculescu and Stephen CameronPattern Recognition, July 2020FPR – Fast Path Risk Algorithm to Evaluate Collision ProbabilityAndrew Blake, Alejandro Bordallo, Kamen Brestnichki, Majd Hawasly, Svetlin Valentinov Penkov, Subramanian Ramamoorthy, Alexandre SilvaIEEE Robotics and Automation Letters ( Volume: 5, Issue: 1, Jan 2020)Scalable FPGA Median Filtering using Multiple Efficient PassesOscar Rahnama, Tommaso Cavallari, Philip H. S. Torr, Stuart GolodetzProceedings of the 2020 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (PFGA’20) (February 2021 Pages 313-313) (poster, non-archival)2019Published papersAnchor Diffusion for Unsupervised Video Object SegmentationZhao Yang*, Qiang Wang*, Luca Bertinetto, Weiming Hu, Song Bai and Philip H.S. TorrInternational Conference on Computer Vision (ICCV), October 2019Correct-by-Construction Advanced Driver Assistance Systems based on a Cognitive ArchitectureFrancisco Eiras, Morteza Lahijanian, and Marta KwiatkowskaProc. IEEE Connected and Automated Vehicles Symposium (IEEE CAVS), 2019Meta-learning with differentiable closed-form solversLuca Bertinetto, João Henriques, Philip H.S. T