Tyre plant material handling: 24 AMRs off the forklift route in 18 weeks

Key facts

Why did this route need automating?

Tyre curing applies heat and pressure to green rubber, so the presses run hot and vent through the cycle. Manual compound handling beside them puts a person in that aisle on every trip, and the number of trips per shift is the number of exposures per shift.

The route also ran two ways at once. Forklifts moved compound slabs from mixing toward the cure hall, and forklift traffic in the aisles set aisle width for the whole building. Hand-pushed trolleys then covered the last leg into the press bays, because a forklift cannot place material at the cadence a cure line asks for. Most trips therefore involved a handover at the drop point, and the waiting happened there rather than on the move.

Four costs followed, none of them specific to this plant:

• Safety. Trolley collisions, manual lifting and heat exposure at the presses produced a recurring pattern of lost-time incidents month on month.
• Line stoppage due to material shortage. A cure line runs to takt, so material arriving outside the window idles a press. Stoppages ran 3 to 4 per week on this route.
• No visibility of material movement. Nobody could say where a given trolley physically was, so work in progress was an estimate rather than a location.
• Labour cost. Peak demand arrived as overtime, and wages on manual material-movement roles have been escalating 30 to 35% year on year, a sector-wide pressure in Indian manufacturing.

What changed on the route?

Every figure above is Projected, from Virya's deployment model as of August 2026, and each will be replaced with a measured value once instrumented.

Operators moved into fleet supervision. The people who ran compound trolleys moved into fleet supervision, quality control, packaging and safety roles, which is more skilled work and it is done away from the presses. The plant did not lose those people; it redeployed them.

Manual trips beside the presses stopped. The trips still happen. A robot makes them.

Aisle width is no longer set by forklift traffic. Once slab movement runs on AMRs, the aisle only has to fit the AMR.

Every trolley is tracked from pickup to drop. Each one is scanned at pickup and tracked to the drop, so work in progress has a location on a screen instead of an estimate from whoever last saw it.

How does Virya's stack run the route?

The parts below are platform capabilities rather than anything built for this plant, so they apply to any process with the same shape: a fixed source, a fixed destination, and a trolley in between.

Fused multi-layer perception. The AMR 50 merges 3D LiDAR, 2D LiDAR, inertial and depth-camera data into one model of the aisle rather than reasoning from a single sensor. That is what makes it an indoor-outdoor AMR in practice: the same platform runs a press aisle and an open yard leg without a change of hardware.

Autonomous docking and hitching. The robot approaches a trolley, resolves its position and orientation, and couples without a person touching it. Virya designs the trolley-side hook against the trolleys a plant already owns, so autonomous hitching works on the existing fleet rather than requiring new trolleys. Unhitching at the drop is the same operation in reverse.

Trolley-oscillation detection. A towed trolley wants to swing, and a swinging trolley either slows the fleet down or hits something. The AMR 50 monitors trailer behaviour and moderates speed and cornering to keep a towed load stable, which is what makes 5,000 kg towing usable in a working aisle rather than only in a straight line.

Custom lane engineering. Routes are laid out as bot-only, mixed, or pedestrian-shared lanes, with zone speed limits set per area rather than per robot. On this deployment that is what handles the press aisle and the shift-changeover crossing differently without writing two different robots.

AMR MES integration, deliberately shallow. The Virya fleet management system runs on a server inside the plant and takes a one-way feed from the plant's MES. Each process has a call screen; when a station needs material an operator presses it, and the fleet manager assigns the robot best placed to go by position, current load and queue. Nothing inside the plant's MES changed, which is what let the fleet go in without an IT project running alongside it.

Battery hot-swap on a 3,000-cycle pack. The standard AMR 50 pack is 48V, 240 Ah, rated 3,000 cycles and charging in about five hours at 50 A. Sites running three shifts can take a hot-swap configuration instead, which takes the charge window out of the shift plan entirely.

Site preparation is floor markings, not construction. Parking and home positions get marked, pickup and drop points are fixed per process, and gangways need width and turning room for a robot towing a load. No floor cutting, no racking change, and no line stopped for installation.

How long does an AMR deployment take in a working tyre plant?

A pilot ran in 6 weeks. The full 24-robot fleet was validated and in place in 18 weeks. Production continued throughout, which is the part most plants ask about first.

High-fidelity 3D LiDAR site mapping is what makes the front of that timeline short: the plant is mapped densely at install rather than surveyed and marked out over weeks. The pilot then proves the route and the takt integration on a small number of units, which is where this kind of deployment either works or does not. The rest of the rollout is a repetition problem rather than a design problem.

How much has Virya's fleet run in total?

Across all sites, Virya's deployed fleet has executed 1,23,502 transport tasks and covered 1,66,686 kilometres autonomously since November 2025.

Both figures are measured rather than modelled, and both are fleet-wide totals across every Virya deployment rather than this plant's numbers.