
By commission
Available by special commission, built by the team that designed it.
Enquire about a commissionIllustrative image
01The experience
Designed and built from first principles. Explore the vehicle, the intelligence and the intellectual property available to automotive partners.
This page is the car, opened up. Turn the original CAD model, switch on its lights, lift the boot and take it apart. Watch what it sees, try the ideas behind our patents, then follow the journey from the first patent filing to public roads.
02Ways to own it
Bring the intelligence developed through Kar-go into the vehicles you build.

By commission
Available by special commission, built by the team that designed it.
Enquire about a commissionIllustrative image

To license
The in-car AI experience, under the carmaker's own name. It never steers, brakes or changes a safety setting.
Visit the siteIllustrative image

To license
The driver camera that sees. Points, not pictures.
Visit the siteIllustrative image
In development
Road perception technology for infrastructure operators and road authorities. In development.
Register interestIllustration
To license
Our own computer: a board we design, with our own system on a chip at its heart, patent pending, and small foundational AIs built from scratch to run on it. It can go into the vehicles you make.
See the boardIllustration
White in the city rain, red on an English country lane, blue on a salt flat at dusk.
Rendered from Kar-go’s CAD in concept colours. Not footage.


From here, its own AI shows you what it sees and explains what we can't.
Drag to rotate
03The machine
Explore the original vehicle geometry. Rotate the model, inspect each system and follow the assembly from bodywork to platform.
Dimensions measured from the Kar-go CAD model.
The lights
Headlights, wheel pod illumination and a programmable canopy display form a single visual language.
The boot
A rear opening provides direct access to the parcel bay. In November 2020, Kar-go delivered medicines from a pharmacy to care homes in Hounslow without human contact.
Active vision
Compact agents direct attention towards relevant parts of a scene, allocating detail where it is needed.
GB 2599364 granted in the UK
Explore active visionUnder the skin
Cameras, our own AI board and the parcel bay are connected through the vehicle nervous system.
Body, wheels, pods and boot: Kar-go CAD. The internal systems are illustrative.
Assembly
An exploded view of the original CAD reveals the relationship between each panel, pod and wheel.
Intellectual property
Developed by our own team. Active vision is granted in the UK, and our four European applications are approved for grant; the rest are filed. Select a marker to explore each principle.
Reassembled
The vehicle is built by special commission. Its intelligence is available for automotive partners to license.
04What it sees
Our active vision approach directs compact agents towards the parts of a scene that matter. Explore how attention shifts between an initial scan, closer inspection and continued observation.
Illustration on a generated film. The boxes come from a standard object detector run on the film; the agents are drawn to show how active vision works, not measured from a vehicle.
05Studies and patents
Three interactive studies explore motion sensing, network evolution and contextual memory. Each demonstrates a principle from our intellectual property portfolio.
Study one · Motion sensing
Explore how radio-based motion sensing can complement a camera when a parked vehicle obscures the view.
GB 2585332 filed · offered as IP to license
An illustration of how it works. Little Feet is real, and designed to go into any vehicle.
Study two · Network evolution
A population of compact networks competes within a set budget. Adjust the available neurons and observe successive generations.
GB 2599121 filed · EP 4217930 approved for grant · US 2023/0368513 · WO 2022/064209
Invented by Dr Aparajit Narayan and William Sachiti, from research at Aberystwyth University with Professor Elio Tuci. Simplified demonstration of the principle.
Study three · Contextual memory
Select a location, sound or word to see related memories surface. Each driver has a separate memory file, held on the vehicle or their phone.
UK application filed December 2023 · GB202318450D0
The driver and the memories are made up for this illustration.
Eighteen published filings in seven families, and the UK filings before them, from 2013. Active vision is granted in the UK; our four European applications are approved for grant. Every number, with its registry link: the record.
Where it started, before Kar-go: William's first patent filings, in 2013, 2014 and 2015, on power and guidance for robots and unmanned aircraft.
A delivery vehicle re-plans its route around obstructions, roadworks and changed deliveries, keeping every stop it has not yet made.
A sensor low on the vehicle reaches under and behind parked cars, and tells a child's moving legs from things that stand still, before a camera can see them. Offered as IP to license.
Small agents decide where to look, how closely and for what, instead of reading every pixel: camera understanding on the computer a car already has.
Granted on 12 April 2023.
European application approved for grant. Inventors: William Sachiti and Dr Aparajit Narayan.
A way to fit compact models to particular hardware: networks tested side by side, the fittest bred into the next generation.
European application approved for grant. Inventors: Dr Aparajit Narayan and William Sachiti.
A computer fitted to the vehicle runs active vision on the vehicle itself, with the cameras and sensors it works with.
European application approved for grant. First-named inventor: Dr Aparajit Narayan.
Fleet and software-defined vehicle design that keeps data to a minimum and has no single point of failure.
European application approved for grant. Available to license.
Cameras on a passing vehicle capture a pothole in stereo, and a neural network measures its size and depth.
An assistant that remembers each driver on board, woken by a place, a sound or a word.
Filed in December 2023.
Every filing, with its registry link and inventors: the record.
06The fleet, RAF Brize Norton, 2021
Trialled with the Royal Air Force at RAF Brize Norton, in a trial part-funded by its Astra programme. Explore the connected vehicle nervous system across six stages, including its response when the command hub is disconnected.
GB 2610855 filed · WO 2023/041930 · EP 4419975 approved for grant · available to license
Fleet and software-defined vehicle design that keeps data to a minimum and has no single point of failure. Our first internal AI began there, in a mobile command hub. Home since 2022 at RAF Neatishead.


07On film
Independent coverage and our own footage of Kar-go, from its London launch to Goodwood and RAF Brize Norton. Illustrative films, made with generative video, come last and are marked.
Each piece is the outlet's own account. We describe them in our own words.
The record: every claim on this site, with its source.
08The journey
The research, vehicles and deployments behind the platform.
2013
Before Kar-go, William filed his first patent application, on power and guidance for robots and unmanned aircraft. Eighteen published patent records followed, in seven families.
Drawn for this page
2016 · Aberystwyth
Robotics before Kar-go: at Aberystwyth University, ask Hugh for a book and it walks you to the shelf. Built with Ariel Ladegaard.
Drawn for this page · archive photograph to come
2017 · The workshop
Aberystwyth University's £10,000 InvEnterPrize, won with Ariel Ladegaard and Aparajit Narayan, and Kar-go's prototype taking shape, designed by our team.
Archive photograph
2017 · Design
An early render of the compartments inside: a bay that opens at the back, at standing height, so nobody has to crouch.
Archive render
2019 · Goodwood
Kar-go is unveiled in the Future Lab at the Goodwood Festival of Speed.
Archive photograph
2019 · Glasgow · London
Meeting people, and the press, in Glasgow and London.
Archive photograph
2020 · Public roads
Kar-go, the UK's first street-legal autonomous vehicle: designed from scratch by our team, and approved by the DVSA as roadworthy for trials on public roads, with a safety driver. That September we filed active vision and training by evolution.
Illustrative film · generative video
Nov 2020 · Hounslow, London
Medicines from a pharmacy to care homes, with no human contact, during the pandemic. Funded by UK Research and Innovation.
Illustrative film · generative video
2021 · RAF Brize Norton
Our vehicles and a hub shared signals both ways through our connected vehicle nervous system; at every checkpoint each vehicle forgot what it didn't need. Athena, our first internal AI, began there, in a mobile command hub.
Archive photograph
Since 2022 · RAF Neatishead, Norfolk
A radar station since 1941. Its Type 84 radar, in service from 1963 to 1993, is a scheduled monument, the last large Cold War air defence radar surviving in England; we taught it to listen, as a research project in sensing.
Drawn for this page
Nov 2022 · Milton Keynes
Trialled at Milton Keynes University Hospital, an NHS hospital, from November 2022.
Drawn for this page · archive photograph to come
April 2023 · Granted
GB 2599364, granted on 12 April 2023, with European, US and international applications behind it.
The patent seal
Dec 2023 · Filed
An application for advanced memory systems for AI: an assistant that remembers each driver, woken by a place, a sound or a word.
Drawn for this page
2026 · For carmakers
What Kar-go taught us, for other people's cars: the in-car AI and Driver Awareness, under the carmaker's own name.
Illustrative image
09The car
Available by special commission, built by the team that designed it. Explore the form, the details and the original vehicle dimensions.






10The platform
Bring the intelligence developed through Kar-go into the vehicles you build. Explore the software, intellectual property and integration routes available to automotive partners.
11Contact
Speak to the team about licensing, vehicle programmes or a special commission.
sales@academyofrobotics.co.ukListening