Gravis launches autonomous earthmoving tech

Construction automation usually gets pitched like a moonshot. New robot. New fleet. New workflow. But dirt work is not software, and contractors do not swap out excavators the way offices swap laptops. That is why Gravis Robotics’ U.S. launch matters. The company is not asking builders to wait for a fully robotic future — it is selling a retrofit path that starts with machine guidance and can step up to autonomy later. (powerprogress.com) ### What did Gravis actually launch? At CONEXPO-CON/AGG 2026 in Las Vegas, Gravis said it was entering the U.S. market commercially and introduced Gravis Copilot, a machine-guidance system for earthmoving equipment. Think excavators first, but the broader pitch is off-road heavy equipment. Copilot runs on a tablet-like interface Gravis calls Slate and sits on top of the company’s retrofit hardware and sensor stack. (powerprogress.com) ### Why is “retrofit” the important word? Because that is the whole wedge. Instead of waiting for OEMs to ship fully autonomous machines at scale, Gravis adds its Rack kit to existing equipment. That kit bundles sensors and onboard compute so a standard excavator can see the site, understand a target dig plan, and execute parts of the job with much less manual intervention. For contractors, that is a much easier buying decision than replacing fleets. (hitachicm.com) ### So what does Copilot do before full autonomy? It gives operators a live digital map of the job. They can upload a CAD file, see target terrain, boundaries, and even buried utilities, then dig or grade without constantly checking paper plans or tape measurements. That sounds incremental, but on a real site the small frictions are the job — stoppi(hitachicm.com)loop. (hitachicm.com) ### Where does autonomy come in? Gravis is using a two-tier model. First, make skilled operators faster and safer. Then automate repetitive tasks like trenching or predetermined excavation paths. In Hitachi’s CONEXPO demo, a ZX135US-7 excavator could be trained by an operator and then switched into autonomous trenching mode within minutes. That is a(hitachicm.com)one operator set the intent and let the machine handle the repeatable part. (hitachicm.com) ### Why is earthmoving such a hard category? Because soil is not uniform. Ground changes. Slopes drift. Machines bounce. Visibility is bad. People walk into work zones. A warehouse robot gets a flat floor and fixed shelves; an excavator gets mud, glare, vibration, and a trench that never looks exactly like the model. Gravis says its system fuses LiD(hitachicm.com)ive heavy equipment both eyes and feel. (powerprogress.com) ### Is this still mostly a demo story? Less than it used to be. Gravis is pointing to live deployments beyond the trade-show booth — autonomous excavation at Manchester Airport, quarry work with Holcim, and integrations across multiple OEM platforms. The company also says its systems can lift throughput by up to 30%, though that number will obviously (powerprogress.com)ion-scale rollout. (powerprogress.com) ### Why should engineers care? Because this is where robotics hits the physical world for real. The stack needs perception, controls, hydraulic integration, human-machine interfaces, safety logic, and real-time decision-making that survives dust and vibration. If this category grows, demand will not just be for AI people in the abstract. It will be for engineers who can make sensors, controls, and safety systems work on unforgiving machines. (powerprogress.com) ### Bottom line? The interesting part is not that Gravis says excavators can drive themselves. Plenty of companies have said versions of that. The interesting part is the rollout model — guidance first, autonomy second, retrofit throughout. If that works, autonomous earthmoving stops being a futuristic fleet story and starts becoming an upgrade path.