A team of leading academics and researchers in AI, simulation, strategy and engagement firms has been awarded £2 million in CCAV funding to advance DeepSafe, the next stage in the commercialisation and deployment of self-driving vehicles.

The dRISK.ai-led consortium of DG Cities, Imperial College London, Claytex and rFpro will unlock a barrier in the supply chain – together, they will develop the simulation-based training needed to train autonomous vehicles (AVs) to handle ‘edge cases’, the rare, unexpected driving scenarios they must be prepared to encounter on the road.

DeepSafe will commercialise ‘sensor real’ edge case data – a simulation of what an actual sensor would detect – together with AV training tools, for release in the UK and internationally after the project. As well as advancing self-driving systems, the grant aims to support innovation in industry, job creation, and investment, building the capacity to develop AV technology in the UK and export it to the rest of the world.

 

Rav Babbra, dRISK.ai: “There are currently bottlenecks in the technology sector that are inhibiting deep-learning using simulation. These are what the DeepSafe consortium aims to resolve. Together, we will enable the performance improvements in AI perception needed to develop safer and more reliable autonomous driving behaviours.”

Pilot studies have demonstrated that AVs can handle the straightforward situations that make up 99% of everyday driving experiences. However, it is more difficult to train vehicles to deal with ‘edge cases’, the rare and unusual events that can happen on the road. Failure to understand an edge case can result in unreliable, unsafe behaviour, such as phantom braking, lane-keeping failure and collisions.

To train AVs to safely navigate these situations, they have to be identified and simulated. Part of the team’s previous project, DRISK focused on engagement, using the public’s driving experiences to crowd source these rare events and build scenarios – a process that also served to raise awareness of the technology and give an insight into consumer barriers to autonomy.

However, such real-life training data is limited. There is a widely recognised need for simulation and synthetic data, but the artificial data currently available is not sufficiently ‘sensor-real’ for AI trained by simulation to improve its perception and decision-making. DeepSafe brings together leaders in the simulation supply chain to resolve these synthetic data issues to enable the successful simulation-based training vital to develop safe, reliable self-driving services.

In addition, DeepSafe will establish the definitive toolchain to realistically represent the dynamics of vulnerable road users, and will conclusively answer ‘how close does simulation have to be?’ to train an ADS (Automated Driving System) to outperform a human driver by an order of magnitude.

Complementing the new data and simulation techniques will be a novel “AI Canvas” – a new kind of software for understanding the weak points of learned systems, backtracking failure modes to their source, and correcting them with new training data. With this complete toolchain, the project hopes to finally unblock the way to safe and economical AV deployments.

Balazs Csuvar, Head of Delivery at DG Cities: “A lot of our focus within the consortium is on user perception – how can we make sure that people feel safe in self-driving cars, how can they be reassured by and trust the work that has gone into the testing. Lane-keeping assist and other ADAS features (in some way precursors to fully autonomous driving) already influence the way we drive, without people actually knowing how well they perform. We will work on understanding how best to communicate to drivers this crucial information and use it to outline how autonomous systems should be benchmarked as well.”

More details here: https://zenzic.io/news/announcing-the-13-successful-cam-supply-chain-uk-projects/

The UK government is awarding almost £42 million to 7 projects through the Centre for Connected and Autonomous Vehicles (CCAV) Commercialising Connected and Automated Mobility (CAM) competition. Industry consortia will match the public grant to around £81 million and will be expected to demonstrate a sustainable commercial service by 2025. dRISK play a pivotal and safety critical role in two of these projects:

Project Cambridge Connector – Greater Cambridge Partnership – £17.4 million to trial on-demand, self-driving taxis, to complement existing transport services in parts of Cambridge.
Multi-Area Connected Automated Mobility – Conigital – £15.2 million to establish a remote driving control hub, to oversee self-driving vehicles operating in Solihull and Coventry, with the NEC and local councils.

Multi-Area Connected Automated Mobility – Conigital
£8.3 million awarded by government, matched by industry to a total £15.2 million. This project looks to establish a self-driving vehicle operation around various parts of the West Midlands, underpinned by a centralised, Remote Monitoring Teleoperation (RMTO) centre. The RMTO centre will be where the project’s self-driving vehicles are monitored and (when required) controlled from, using 5G connectivity. The project aims to make self-driving vehicle operations commercially viable, and offset current technology and driver costs.
Project partners include: National Exhibition Centre Ltd, Direct Line Group, Coventry City Council, Solihull Metropolitan Borough Council, University of Warwick, Coventry University, dRISK, IPG Automotive and West Midlands Combined Authority

Project Cambridge Connector – Greater Cambridge Partnership
£8.7 million awarded by government, matched by industry to a total £17.4 million. The Cambridge project will pilot on-demand self-driving vehicles. Up to 13 electric vehicles will provide passenger services that integrate with existing transport services within Cambridge across two sites: Cambridge University’s West Cambridge Campus and the Cambridge Biomedical Campus.
The lead partner Greater Cambridge Partnership is the local delivery body for a City Deal with central government, and working in partnership with Cambridgeshire County Council, Cambridge City Council, South Cambridgeshire District Council and University of Cambridge.
Project partners include: dRISK, Stagecoach East, IPG Automotive UK, Conigital Ltd and Gamma Energy as well as the Greater Cambridge Partnership.

UK Business Secretary Grant Shapps said:
“In just a few years’ time, the business of self-driving vehicles could add tens of billions to our economy and create tens of thousands of jobs across the UK. This is a massive opportunity to drive forward our priority to grow the economy, which we are determined to seize”.
he went on to say
“The support we are providing today will help our transport and technology pioneers steal a march on the global competition, by turning their bright ideas into market-ready products sooner than anyone else”.

UK Transport Secretary Mark Harper said:
“Self-driving vehicles including buses will positively transform people’s everyday lives – making it easier to get around, access vital services and improve regional connectivity”.
“We’re supporting and investing in the safe rollout of this incredible technology to help maximise its full potential, while also creating skilled jobs and boosting growth in this important sector”.

CEO and Founder of dRISK, Chess Stetson said : ” Following the successful delivery of the World’s first true driving test for Self-driving cars to the UK government in September 2022, dRISK are honoured and eager to use this technology to facilitate the next generation of truly safe, truly performant Autonomous Vehicle deployments for the public good”.

Full press release here.

dRISK are delighted to have been featured in the Business Cloud Transport Tech top 50 innovators of 2022.

The company has been recognised as creating innovative transport technology solutions for our work testing, training and validating on Edge cases gathered from all over the globe and it is a true recognition of the work the team have put in to scale our products and services in response to customer demand in both the United States, but also in the UK, mainland Europe and further afield via our outreach missions.

We are proud to have been selected amongst companies who we know and work with on a regular basis within the UK connected and autonomous vehicle eco-system.

The ∂RISK team look forward to continuing successes within the UK as well in the USA, via our US office in Pasadena (CA), with new product offerings, solutions and thought leadership in the Autonomous vehicle safety space, to be released soon.

dRISK were awarded Edge innovator of the year 2022 at the Edge Computing World conference in Santa Clara, Silicon Valley. Our Transport Systems specialist, Kiran Jesudasan collected the award in person and gave the audience a view of the work dRISK are pushing toward in the relatively new world of Edge computing. Kiran described how dRISK specialises in collecting Edge cases, which are the countless risky vehicle scenarios which are individually unlikely, but together make up all the risk. These are used by the company to test, train and validate autonomous vehicles. dRISK collect millions of hours of data from many sources from all over the world. One of our largest data sources is CCTV. The data we collect is mostly made up of perfectly normal driving behaviors and does not contain the edge cases we are interested in. At this time we collect all the data and store this in our servers prior to interrogating it to find the valuable edge cases. Money and time is lost in sending the non-valuable data back and storing this. Kiran spoke about how Edge computing would mean that CCTV cameras themselves would be upgraded to have capabilities themselves to decide what was an Edge case and what was not, at source. In this way, only the valuable data would be sent back to dRISK and we would save on time and financial burden, allowing us to collect yet more data and edge cases to make self driving vehicles safer.

Edge computing means that powerful decision making processes are no longer limited to servers in offices, but can now be deployed where they are needed and useful. dRISK are working with suppliers such as Nvidia to develop Edge computing capabilities, which in this case will allow us to collect more accidents and near misses making future self driving vehicles safer and thus, commercially viable.

CEO & Founder of dRISK, Dr Chess Stetson was interviewed by Edge Computing World and the article can be seen here.

Tell us a bit about yourself – what led you to get involved in the edge computing market and dRISK

My background is in computational neuroscience, and I spent a long time realizing that most of the value in data was in the unexpected places or edge cases. dRISK builds tools for training and testing autonomous vehicles on edge cases, and perhaps ironically does so with a Knowledge Graph called dRISK Edge, which both stores knowledge, edge cases, and works on a distributed database, all of which have have either a semantic or technical relationship to edge computing. But in the strictest sense of edge computing, our heavy use is coming up, as the object detection feeding our knowledge graph starts to run on edge devices rather than centrally. Moreover, the advent of AVs will see more V2V, X2V and V2X communication requiring super low-latency, which will move decisionmaking to the edge.

What is it you & your company are uniquely bringing to the edge market?

We are bringing a dramatically new way to train and test autnomous vehicles on edge cases and, as I say above, it’s not just window dressing to say that all AVs will have an edge computing component in the near future.

Tell us more about the company, what’s your advantages compared to others on the market, who is involved, and what are your major milestones so far?

We are by far the most comprehensive resource for training and testing on edge cases, and our customers enjoy a 6x or greater improvement in safety performance.

How do you see the edge market developing over the next few years?

I don’t think it’s controversial to say that centralized heavy compute is expensive and subject to burdensome latency. That will change as a large variety of distributed computing resources take over the load where it’s needed.

How do you see the edge market developing over the next few years?

We focus on AVs, so I’ll keep my answers contained there. See above — V2X, V2V, X2V all mean edge computing. But we also work on large distributed graph databases, which are ideal for distributed computing, and it’s easy to see how edge compute could take graph computing all the way to the periphery of the computing process with ease.

….in NVIDIA DRIVE Sim

The future of autonomous vehicles (AVS) relies on simulation. Equally important is knowing what scenarios to simulate. Edge cases are high-risk scenarios, each of which is individually unlikely, but collectively, can make up all the risk of daily driving. Edge cases can be anything from a child following a ball into the street to a head-on collision with a car suddenly veering into your lane.

 

A dRISK edge case detailing a dangerous passing maneuver from an oncoming vehicle recorded from a dRISK exclusive source, and replicated in NVIDIA DRIVE Sim with some domain-specific modifications.

 

AV developers will soon be able to test and retrain their AV stacks on these and many more edge cases using the NVIDIA DRIVE Sim platform. DRIVE Sim, which is built on NVIDIA Omniverse, is an end-to-end platform architected from the ground up to run large-scale, physically accurate multi-sensor simulation delivering a safe, scalable and cost-effective way to bring self-driving vehicles to the roads. DRIVE Sim, already shown to be able to closely match real-life scenarios, will now further recreate dRISK’s knowledge graph of edge cases. Developers testing their AV stacks with DRIVE Sim will be able to validate and retrain their own parameters on the world’s trickiest scenarios.

Following a major grant from the UK government to build the ultimate driver’s test for self-driving cars, dRISK has spent years mapping out the landscape of edge cases from a huge heterogeneity of sources – from CCTV to accident reporting, to first-principles failure mode analysis led by NASA engineers.
Testing, validating and retraining on edge cases before deployment can prevent most of the accidents autonomous vehicles are experiencing right now. As dRISK demonstrated at NVIDIA GTC 2021, developers can achieve a 6x performance improvement on detecting high-risk scenarios by retraining on edge cases, while continuing to get the “center cases” (normal driving) for free.

 

Aggregation methods for dRISK’s edge cases from real-life data, clustered and assembled into a knowledge graph, a data structure flexible enough to capture the heterogeneity of everything that can go wrong on the road.

 

NVIDIA DRIVE Sim has the ingredients to achieve a major goal in AV development—to fully test, validate and even retrain and improve these vehicles in simulation. Chess Stetson, CEO of dRISK, says, “NVIDIA DRIVE Sim is the simulation platform we’ve all been waiting for, which can fully resolve the kind of depth we need to do physically accurate simulations in real time.”

NVIDIA is enabling the next generation of software-defined autonomous systems to OEMs.

One of the most important reasons for physically accurate simulations is to understand the AV perception. failures that would cause an oncoming collision to be missed. In the following clip, DRIVE Sim replicates a T-bone scenario from dRISK’s knowledge graph, complete with partial occlusions. DRIVE Sim makes the hazardous vehicle obvious to the human eye, as it would be in real life. But it’s hard for a standard computer-based object detection system to pick up (in this case, Mask RCNN).
Stetson notes, “The human visual system, evolved over millions of years, is highly adapted to avoid things that are going to hurt or kill you, which it does using extensive motion cues.”

This is in stark contrast to current object detection, which is overwhelmingly based on single-frame object localization and classification, rather than the kind of motion detection that could allow for quick reactions necessary to avoid this kind of T-bone scenario. But as AVS have more exposure to edge cases in a full sensor-real simulation platform, they will indeed be able to develop the capabilities to handle these kinds of high-risk, fast-reaction-time scenarios.

 

An upcoming transverse collision, replicated in high fidelity in NVIDIA DRIVE Sim. This kind of edge case is critical for testing AV reactions to high-risk scenarios in DRIVE Sim’s real-time sensor environment. Note that in this case, a common object detection framework (Mask R-CNN) misses the behavior most important to risk awareness in the scene, but that DRIVE Sim allows for the possibility of retraining AV perception to improve.

dRISK’s mission is to provide the largest, most diverse and realistic scenario database for testing autonomous vehicles. dRISK has been building this testing suite from a massive set of real-world data. Test sets are optimized for specific tasks, such as failure mode identification and curriculum learning for efficient Al training. dRISK and its customers have shown 6-20x improvements in the ability to recognize and contend with high-risk scenarios, though never yet in a simulation environment as capable as NVIDIA DRIVE Sim.

Working relationship:
NVIDIA has licensed dRISK’s scenario data derived from real life, which is organized to be evenly distributed across the landscape of risk, for NVIDIA’S AV development uses and to complement its scenario provision efforts.

As a DRIVE Sim ecosystem partner, dRISK is making its scenario knowledge graph and tools for exploring the space of center and edge cases available to any DRIVE Sim customer.

Capabilities and Goals:
dRISK with DRIVE Sim can efficiently:
– Replicate real-life edge cases, sourced from real-life data, at large scale
– Find failure modes in simulation before vehicles are on the road
– Retrain object detection to have improved performance detecting sources of high risk

In addition to delivering high-risk, true-to-life scenarios for NVIDIA’s use in validating AV stacks against edge cases, dRISK is also exploring the capacity to deliver these capabilities to the wider AV space via a combined use of dRISK’s edge case manifold and NVIDIA AV tools.

Authors: dRISK Team (Chess Stetson, Ph.D., Federico Arenas Lopez, Lorenzo Niccolini, Nils Goldbeck, Ph.D.)
Other contributors: Philipp Hermann, Wael Elhaddad, Carène Kamel

How can simulation be used to determine where ADAS features can be improved? The prestigious Institute of engineering and technology (IET) contacted dRISK to ask us help them test Advanced Driver Assistance Systems (ADAS). ADAS features are becoming more common place on the new vehicles, these include Automatic Emergancy Braking (AEB), Automatic Lane Keeping System (AKLS) and Automatic Cruise Control (ACC). ADAS functionality is seen as the first baby steps towards ever increasing driver aid, culminating in fully autonomous vehicles.

We demonstrated our latest set (test two after our Tesla spoofing rig experiment which can be seen here) of testing to see if these driver assistance systems can be relied on. We pressure tested the ADAS functionality on a SAE level 2 autonomous vehicle’s optical perception system by presenting it with challenging situations that could happen in real life.

Do these systems help the driver? Should vehicles equipped with these features cost less to insure as they might be less likely to be involved in an accident? Can the driver be reliant on these features? Watch the video below:

Testing was carried out on a private road in September 2021. No one was hurt in the experiment.

…to central London

dRISK’s London team is growing again and has moved to new offices to central London. Following the successful closure of our Super Seed funding round with Foresight Williams and Stantec, we have been growing our engineering team as a means to scale up and also develop exciting new functionality within the business. To accommodate our larger team, we have moved from our offices in East London to a more centrally located office in the heart of London’s vibrant Holborn.

Katarina Cimesa joins us as an Artificial Intelligence Engineer who has worked at FeatureSpace and IBM. Katarina brings a wealth of experience in inference, especially machine learning and information theory, together with contemporary software technologies for big data.

We are also over the moon to invite our now-ex-Spacex engineer James Kelly as dRISK’s VP of Engineering! James’ career has touched nearly all sides of the GPU, from computer graphics to GPU-accelerated browser interfaces, and he is the perfect mind to help us represent both what the computer eye needs to see of simulated edge cases, and what the human eye needs to see of embedding projections.

In conjunction with the University of Cambridge, this year dRISK are running a summer internship program. We have taken on three summer interns to help us with our mission to make AV’s commercially viable by making them responsive and safe, no matter what the road conditions might be. Ananya Jeyappragash, Isaac Brant and Mohammed Yaseen (left to right), all join us during August and September and we hope to be able to make great use of their combined skillset in relation to dRISK’s core mission.

A very warm welcome to our new starters and interns.

dRISK have been chosen as one of the eleven startups to be selected for the latest batch of companies who will be working with the MobilityXlab and expert partners. MobilityXlab offers startups a fast track into a potential strategic partnership with their seven industry partners; Volvo group, Volvo Cars, Zenseact, CEVT, Ericsson, Polestar and Veoneer. We hope that our work with these world leading partners will ensure that autonomous vehicles become a commercial viability.

MobilityXlab is a collaboration hub founded in 2017 by global companies to create and develop new innovations within future mobility — with each other and with startups. Our seven partners are CEVT, Ericsson, Polestar, Veoneer, Volvo Cars, Volvo Group, and Zenseact. Lindholmen Science Parks is the host organization. MobilityXlab is also supported by Region Vastra Gétaland and Vinnova, Sweden’s Innovation Agency. Over the first five years, MobilityXlab has seen startups applying from 50 countries. The collaboration platform has resulted in 75 proof of concepts and 12 accelerations, in the form of commercial contracts or partnerships.

We are looking forward to working closely with the partners to help them to test, train and validate autonomous vehicle perception stacks enabling autonomous vehicles to deal with any scenario and every variation.
This will mean that we can make AVs collision-free and versatile enough to handle anything the real world throws at them.

Foresight Williams Technology leads the £1.7m super-seed round funding which will support further expansion of the business.

26 May 2022, London: Foresight Group (“Foresight”), a listed private equity and infrastructure investment manager, is pleased to announce a £1 million investment into dRISK (“the Company”), an autonomous vehicle (“AV”) testing and training company. The investment was led by Foresight Williams Technology (“FWT”), a joint venture between Foresight and Williams Advanced Engineering (“WAE”).

Founded in 2014, by CEO Chess Stetson, dRISK is an AI company which promises to revolutionise AV efficiency and safety by training AVs to avoid unexpected real-world “edge case” scenarios. Collectively, edge cases account for most of the serious risk on the road but they are individually unlikely; making them hard to catalogue. AV developers are trying to overcome this by collecting data by driving multiple miles. It is estimated this approach will take decades before AVs become commercially viable.

dRISK’s solution exposes AVs to edge cases in a different way and in a fraction of the time. It assembles the evidence of what has caused accidents and near misses – both by human-driven vehicles and AV/Advanced Driver Assistance System or “ADAS” enabled vehicles – and reconstructs conditions in simulation with a combination of real and synthesized data.

dRISK‘s core technology – with four patents granted and two pending – uses networks of data to store, visualise, and classify “unknown unknowns” in a way that is optimised for AV training systems to absorb. dRISK’s statistical robustness and auditability has allowed it to win the largest grant from the UK’s Centre for Connected and Autonomous Vehicles; enabling dRISK to build the ultimate driver’s test for self-driving cars. dRISK already counts multiple leading AV developers among its current customer base which is growing quickly.

Global design firm, Stantec, alongside previous investors – Okapi Venture Capital, Netsu Equity, SaaS Venture Capital, Poetic Partners and Mt. Wilson Ventures – participated in the round.

The global market for AI training data alone is currently valued at $1.5 billion. FWT’s investment will accelerate the development and growth of dRISK as a central technology for AVs.

Commenting on the investment, Hugh Minnock, Senior Investment Manager at Foresight, said: “Foresight Williams Technology is delighted to invest in dRISK, a hugely exciting company that will hasten the Autonomous Vehicle revolution and the decarbonised future we need to follow.”

Chess Stetson, Chief Executive of dRISK, added: “This investment will allow dRISK to accelerate our business plan and achieve our goal of introducing the dRISK product to the majority of the AV industry. And in turn, dramatically accelerate the development of safe and useable autonomous vehicles. We thank Foresight and Williams Advanced Engineering for their support.”

Matthew Burke, Head of Technology Ventures at WAE, commented: “dRISK’s proprietary technology has the potential to accelerate the development of automated driving control systems and overcome one of the main barriers to self-driving: the identification of edge cases. We are delighted that FWT has made this investment and we expect to leverage our knowledge of the automotive industry to increase uptake of dRISK’s technology.”

The society of Automotive Engineers or SAE as they are better known have published a paper jointly authored by ∂RISK and Hitachi Europe. The paper is titled “Virtual Verification of Decision Making and Motion Planning Functionalities for Automated Vehicles in Urban Edge Case Scenarios” [2022-01-0841]

Despite recent advancements in Automated Driving Systems (ADS), the deployment of such systems in dense urban environments still faces a challenging problem: in comparison to motorway or rural driving, urban environments contain a significantly greater number of traffic participants. This makes it difficult to verify the Safety Of The Intended Functionality (SOTIF) across the entire Operational Design Domain (ODD). One approach to solve this problem is to virtually evaluate and verify the safety of the ADS using simulation tools. Whereas traditionally simulated verification has attempted to replicate normal driving conditions, it is possible to achieve superior safety performance by exposing the ADS to more high-risk scenarios than it would otherwise see in the real world. This paper presents the virtual verification process for decision making and motion planning functionalities in urban high-risk edge case scenarios. At the outset of this study a novel data-driven methodology is used to define numerous urban driving edge cases based on real-world road traffic collisions and near-misses, rather than merely everyday driving. These edge cases are then translated into driving scenarios based on the OpenSCENARIO 1.0 standard and then simulated in batches using the open-source CARLA simulator. The simulator is connected to a decision making and motion planning functionality responsible for controlling the state of the ego-vehicle. The efficacy of this edge-case-based virtual verification pipeline is demonstrated with practical examples where the performance of the functionality in urban driving edge cases is iteratively improved to meet the safety requirements.

Thanks to Lorenzo and Shona from ∂RISK and to Ioannis Souflas, Ludovico Lazzeretti, Alireza Ahrabian from Hitachi Europe for allowing this great research opportunity.
The paper can be accessed from the SAE website here.