Meet Creation Labs – the startup that is set to transform the transport industry with the help of AI
How was the idea behind your startup born? Present your team and let our readers know how you all came together.
Creation Labs was founded by Jakub Langr and Dr. James Hennessey in 2020 with a vision to make road logistics greener, safer and more efficient. They are joined by a team of AI, semiconductor & automotive experts, taking up both engineering and product roles, having previously worked at Amazon, Graphcore, Lyft and Jaguar Land Rover.
The idea for Creation Labs came as a result of Jakub’s family ties and their trucking operation. We knew from this experience and from speaking with fleet managers how important the bottom line is, how much of an operating cost fuel is for a road logistics company, and the pressure they are under to be more eco-friendly.
Can you please present your solution? What makes your solution stand out from the rest?
Our solution is based on the insight that driver behaviour has a huge impact on energy efficiency – there can be a 30% difference between most and least fuel efficient drivers. Our first product is an Advanced Driver Assistance System (ADAS) for diesel powered trucks. We assembled the hardware, integrated it with a vehicle and developed AI software to assist drivers be more efficient.
Our device provides ‘SAE Level-2’ autonomous driving capabilities assisting the driver by using cameras and AI to take control of the vehicle’s braking, steering and acceleration whilst under driver’s supervision. The AI technology has been developed to be fuel-efficient, anticipating other road users’ intent thus helping to reduce fuel-bills and CO2 emissions. Experiments we carried out show that our ADAS can reduce fuel consumption by 10.8% compared to the average driver.
Our solution’s uniqueness comes from two areas: (i) it is specifically optimised to control the vehicle energy efficiently and (ii) we have developed AI algorithms to run in real-time on a low powered processing unit. Unlike other AV technology that requires powerful and expensive GPUs, we use much cheaper hardware.
Which Data Provider do you collaborate with and how has your experience been? What made you choose them and their challenge?
We began our EDI journey initially focused on building a simulation platform (‘GAN-Sim’) to help autonomous vehicle companies train their AI algorithms. We collaborated with our Data Provider Smart Mobility Living Lab (SMLL), a subsidiary of Transport Research Laboratory and testbed provider for AVs. SMLL were able to provide us with a 3D replica of Greenwich, including roads and lanes. We then imported these assets directly into our simulation engine.
This collaboration gave us the foundation to build a unique, more realistic simulation engine. We were able to make use of a combination of computer graphics and a new technique called domain adaptation so our ‘GAN-Sim’ generates perceptually real edge cases. James Long, Head of Technical Consulting at SMLL stated that “the emphasis on the value of simulation is potentially even more relevant given global challenges relating to COVID-19”.
You have come a long way since the start of the EDI incubation programme. Can you tell us more about the evolution/traction of your solution, company and team during the programme?
After carrying out additional market research and speaking with AV companies, we concluded that market demand for the simulation engine was limited. This was mainly due to most AV companies already having their own in-house simulation engines to run experiments on. We, therefore, decided to pivot and make use of the ‘GAN-Sim’ platform ourselves, building automotive-based products to hasten the transition to a lower-carbon economy and pursue a bigger commercial opportunity.
As such, we have built a fuel-efficient, AI-powered ADAS for trucks in Europe. The simulation engine now helps us to: generate perceptually real edge cases for our own AI algorithms; train our ADAS/AV systems using our GAN-Sim approach, which is as effective as training with real data; and scale our self-driving capabilities in a cost effective way (since we do not need to manually drive billions of miles).
We are in extended discussions with OEMs / Tier-1 automotive suppliers and already have signed Letters of Intent from leading UK logistics operators. As trials are planned this year, we have already grown the team to five members and intend to expand further to a total of eight before the end of the year.
What challenges did you face and what lessons did you learn?
Selling a highly technical product to B2B customers can often be a lengthy process due to the number of stakeholders and approval layers involved. Thus, in order to provide investors with evidence of product validation, we were able to obtain commitments from a number of key road logistics companies to sign Letters of Intent as part of our commercial discussions.
Like all companies, we had to adapt to the constraints which resulted from the COVID-19 outbreak. Externally, the pandemic placed a huge strain on our initial target market, road logistics operators. As they were forced to operate their vehicle fleets with minimal downtime, identifying partners with the available capacity to run trial deployments of our technology took longer than anticipated.
What impact has your solution achieved/ you are planning for it to achieve?
As we spoke to more logistics firms, it became clear that what was halting a switch to electric vehicles was the limited range of vehicles for payloads they require. We knew our solution had to help existing diesel-based vehicles be more fuel efficient and also offer a path to extend the range of battery electric vehicles, which will likely become the dominant vehicle platform over the next 10 years.
We have developed a prototype ADAS product using two diesel-based vehicles (passenger and commercial). The internal study of our ADAS fuel savings demonstrates that it can reduce diesel consumption by 10.8%. This evidence, we believe, indicates that our ADAS will also be able to extend the range of electric vehicles, although this work has yet to begin. As such, we are in discussions with battery electric OEMs to assess whether they would be willing to pilot our ADAS technology in their vehicle.
According to your experience, what is the secret behind a successful data driven startup?
Quality Data – majority of data driven startups have been successful by using quality data rather than optimising to source copious amounts of data. ‘Quality data’ can be defined by several characteristics but in our case, we focused on utilising data that was accurate/precise, reliable/consistent, complete/comprehensive. This approach is important given our ADAS product has been built for public roads so has minimal tolerance for error.
Focused KPIs – establishing key commercial and technology-based performance indicators helped us to visualise what the team’s version of success looked like and how to easily measure it. Setting correct and robust KPIs, which excluded vanity metrics, helped us to prioritise what we needed to do to move the needle.
A/B Experiments – it’s one thing telling customers that we can make their vehicles more fuel efficient, it’s another thing showing them our actual experiments and original data points to support our claims.
What’s next for you? Are you looking for partnerships, a new round of investment, new piloting or something else?
We are soon to finalise our seed round of investment. The funding will help us achieve our ambitions of using self-driving technology to optimise for energy efficiency via a Level-2 computer vision-based ADAS and gradually build up to private road Level-4 automation. We will continue discussions with OEMs/Tier-1s who intend to licence our AI software for use in fuel/electric vehicles, with private road testing planned for May 2021. Moreover, we are already in the process of setting up trials with leading global logistics companies and are looking at deploying solutions with more companies to prove out fuel savings across different conditions. The product can pay for itself and allow us to gather safety data on European roads.
How would you describe your experience with EDI?
EDI programme is structured in three core stages: explore, experiment, evolve. As startups are assessed at the end of each stage to determine whether they qualify for the next stage, this encouraged us to remain focused and prioritise hitting product/market fit.
The EDI financing enabled us to acquire hardware and build a working prototype of our AI-powered ADAS solution using both commercial and passenger vehicles. This meant we could collect live driving data to train our AI algorithms optimised for fuel efficiency.
The excellent training we received from EDI coaches ensured we took a more structured and objective approach to assessing our progress. Putting in place commercial and technical KPIs, and regularly reviewing them with our coaches, motivated us to succeed.
