INFRASTRUCTURE-INDEPENDENT NAVIGATION.
NAV-KIT gives existing off-road and autonomous vehicles the ability to orient themselves in unstructured terrain. Onboard computer vision builds a continuously updated spatial representation of the environment. Combined with inertial sensing, vehicle data and GNSS when trusted, NAV-KIT provides the foundation for resilient navigation across forestry, mining, emergency response, remote industry and defence. SEE. ORIENT. MOVE.
EXPLORE NAV-KIT→CONTINUOUS REMOTE DRIVING DOESN'T SCALE.
Many remotely operated vehicles still require a person to continuously watch a video stream and control steering, throttle and braking. This makes every additional vehicle dependent on additional human attention — and places the real-time driving loop across a communications link.
MORE VEHICLES → MORE OPERATORS
MORE SPEED → LESS TOLERANCE FOR LATENCY
- More vehicles require more operators.
- More speed requires less latency.
- Communications may be intermittent, degraded or denied.
- Operators need to supervise outcomes — not steer around every tree.
NAV-KIT MOVES THE IMMEDIATE DRIVING PROBLEM FROM THE NETWORK TO THE MACHINE.
KNOW WHERE YOU ARE FROM WHAT YOU SEE.
A human moving through a forest does not navigate from coordinates alone. We read the terrain. We recognise slopes, tracks, openings and structures. We estimate our own movement and continuously compare what we see with our understanding of the environment. NAV-KIT applies the same principle to a machine. Computer vision, inertial sensing, odometry and vehicle data are not separate inputs into a generic sensor-fusion box. They form an integrated understanding of where the vehicle is, how it has moved, and what surrounds it. GNSS is another source of information. Not the entire answer.
GNSS IS ANOTHER SOURCE OF INFORMATION. NOT THE ENTIRE ANSWER.
BUILD THE WORLD AS YOU MOVE THROUGH IT.
NAV-KIT uses computer vision to continuously create and update a three-dimensional representation of the vehicle's surroundings.
Terrain geometry, visual structure and the vehicle's own movement become part of a local spatial model. NAV-KIT can then relate new observations to that model to understand how the vehicle has moved and where it is relative to the world around it.
THE MACHINE OBSERVES AND RECONSTRUCTS THE WORLD, THEN USES OTHER SIGNALS TO STRENGTHEN AND CONSTRAIN THAT UNDERSTANDING.
Supporting — feeds into localization
FROM TELEOPERATION TO SUPERVISED AUTONOMY.
FOUR QUESTIONS BEFORE THE MACHINE MOVES.
The onboard loop resolves four problems in sequence on every cycle. Together they determine whether the vehicle can proceed, and how.
SEE → ORIENT → UNDERSTAND → PLAN → MOVE → SEE AGAIN.
HUMAN IN THE LOOP. NOT IN THE DRIVING LOOP.
NAV-KIT changes the relationship between operator and machine. Instead of continuously steering the vehicle, the human provides intent — a destination, route or task. NAV-KIT handles the immediate problems of orientation, terrain understanding, path selection and vehicle movement onboard. The operator remains in control and can supervise, reroute, stop or intervene when required.
Intent goes down. Understanding comes back. Driving stays onboard.
MOVEMENT WITHOUT DEPENDING ON THE VIDEO LINK.
Continuous teleoperation places the driving loop across the communications network. At higher speeds, latency, degraded frame rates and intermittent connectivity increasingly constrain remote driving. NAV-KIT moves perception, orientation, path planning and immediate vehicle control onboard. The network carries intent. It doesn't have to carry every turn.
The driving loop runs over the network. Every metre requires a connected human.
The driving loop moves onboard. The human supervises the outcome.
THE NETWORK CARRIES INTENT. IT DOESN'T HAVE TO CARRY EVERY TURN.
BUILT FOR WHERE ROADS END.
Road autonomy benefits from structure: lanes, signs, predictable surfaces and well-mapped environments. NAV-KIT is being developed for places where those assumptions disappear. Forest tracks. Mud. Snow. Vegetation. Ditches. Rubble. Slopes. Unmarked terrain. The key technical distinction is traversability. A bush, ditch, slope or muddy area is not simply an object. Whether it can be crossed depends on geometry, terrain type, surface condition, vehicle width, clearance and mobility characteristics. NAV-KIT ultimately needs to understand terrain relative to the machine attempting to cross it.
IS THERE AN OBJECT?
CAN THIS VEHICLE DRIVE THERE?
AUTONOMY IS A PROGRESSION.
HUMAN ATTENTION PER VEHICLE
NAV-KIT is being developed as a progression. Not all stages have been achieved.
PART OF A LARGER OPERATIONAL SYSTEM.
NAV-KIT first understands where it is and what surrounds it — then determines how to move through that world. MISSION BRAIN: Where should the machine go, and why? NAV-KIT: How does the machine get there? NAV-KIT works as an independent navigation layer and as part of the wider HELM architecture.
NAV-KIT handles local spatial perception; AMI understands the mission implications of those observations.
Explore AMI→FROM REMOTE CONTROL TO AUTONOMOUS MOBILITY.
NAV-KIT is being developed for environments where terrain, positioning and communications cannot be assumed. Give the machine intent. Move the immediate driving problem onboard. Keep the human in command.
DISCUSS NAV-KIT→