AD1 Indoor L4 Kit
The AD1 Indoor L4 Kit is a self-driving retrofit kit, comprising a sensor head, SoC2 compute box, and drive-by-wire actuation interface, designed to convert existing forklifts and tugs into closed-loop autonomous vehicles. Because it operates inside private sites exempt from automotive homologation, it allows DeepGrid to ship commercial Level-4 autonomy years before public-road deployment while serving as the technical proof point for the road roadmap.
- Listed price
- ₹1,00,000
- FY2032 revenue
- ₹42 Cr
- Share of plan
- 3.7%
- Gross margin
- 89%
- First revenue
- FY2028
- Volume plan
- 4,200 @ FY32
From input to outcome
Proposed product architecture. The scene illustrates its application, not a shipped kit.
Cameras and depth sensing read the aisle, load and people
SoC2 localises the vehicle against a site map it holds onboard
Planner picks a path and a speed for the geofenced route
Drive-by-wire interface issues steer, throttle and brake
Drag to rotate. Conceptual layout, not a mask.
On the die
The case in brief
- AD1 deploys closed-loop Level-4 autonomy on SoC2 years before public-road legality by retrofitting existing indoor industrial fleets exempt from automotive homologation.
- The kit generates ₹42.0 Cr at 89% gross margin by FY2032 as a management projection, acting as the live technical proof point for the road autonomy roadmap.
- Commercial scaling is governed by field services throughput for site mapping and commissioning rather than silicon production constraints.
What it is used for
Warehouse and Dark-Store Intralogistics
- Buyer
- Warehouse and industrial operators
- The job
- Operators need to automate repetitive internal material movement, including inbound put-away and pick runs across busy floors, using forklifts and tugs the site already owns.
- What DeepGrid supplies
- DeepGrid supplies a retrofit kit featuring camera and depth sensing, an onboard SoC2 compute box, and drive-by-wire actuation that executes closed-loop steering, throttle, and braking with auto-rerouting.
Material-Handling System Integration
- Buyer
- Material-handling integrators
- The job
- Integrators require turnkey autonomous drive-by-wire and perception packages to automate customer-owned indoor utility fleets without waiting for public-road homologation frameworks.
- What DeepGrid supplies
- DeepGrid provides the complete sensor suite, 24 m LiDAR and camera fusion pipeline, site mapping tooling, and ±5 cm obstacle stop actuation to operate reliably within geofenced indoor zones.
The story
A self-driving retrofit kit bringing closed-loop Level-4 autonomy to existing indoor utility vehicles.
AD1 converts existing forklifts and tugs into autonomous systems that never leave private industrial buildings. The retrofit package combines camera and depth sensing, an SoC2 compute box, and a drive-by-wire actuation interface at a unit price of ₹1.00 L. It allows DeepGrid to ship operational Level-4 behaviour years before public-road legality using the exact silicon architecture planned for automotive deployment.
- ₹1.00 LUnit price
- SystemsSegment
- IndoorOperating domain
- FY2028First revenue
- Camera and depth sensing headSensor component
- Drive-by-wire interfaceActuation component
A closed-loop signal chain executes sensing, onboard compute, planning, and drive-by-wire actuation.
Depth sensing and cameras fuse into a single bird's-eye occupancy view, monitoring aisles, cargo, and pedestrians. The SoC2 compute box localises the vehicle against an onboard site map, while an integrated planner establishes paths and speeds up to 1.5 m/s within a moving safety zone. The drive-by-wire interface executes direct steer, throttle, and brake commands, operating the vehicle entirely without a driver.
- 24 m at 5 HzLiDAR specification
- 57°Camera field of view
- Up to 1.5 m/sClosed-loop speed
- Closed-loop steer, throttle, brakeDrive-by-wire actuation
- Site mapping and commissioningOperational dependency
Warehouse and industrial operators deploy indoor autonomy under purely commercial procurement decisions.
Because private industrial sites operate outside automotive homologation frameworks, fleet automation remains an immediate commercial decision rather than a regulatory process. DeepGrid reaches site operators automating repetitive internal movements both directly and via material-handling integrators. The deployment addresses the warehouse and industrial autonomy segment within the broader $4.8B edge-AI market.
- Warehouse and industrial operatorsTarget buyer
- Direct and handling integratorsDistribution channel
- Exempt from homologationCommercial trigger
- $4.8B edge-AI poolMarket demand pool
- 4,200 units (management projection)FY2032 volume
AD1 delivers high-margin systems revenue while validating the silicon roadmap for public-road autonomy.
AD1 scales to ₹42.0 Cr in FY2032, representing 3.7% of group revenue at an 89% gross margin under management projections. While modest in top-line share, it validates the production perception stack and decision layer on SoC2 in an environment where closed-loop control is legally viable today. Growth is gated by services capacity for site mapping and commissioning rather than silicon supply.
- 4,200 units (management projection)FY2032 volume
- ₹42.0 Cr (management projection)FY2032 revenue
- 3.7% (management projection)Share of group
- 89% (management projection)Gross margin
- Road roadmap proof pointPortfolio role
- Services capacity throughputScaling dependency
- ₹0 CrFY270 units
- ₹0.2 CrFY2820 units
- ₹4 CrFY29400 units
- ₹10 CrFY301,000 units
- ₹22 CrFY312,200 units
- ₹42 CrFY324,200 units
A warehouse simulator validates the production perception stack and real-world hazard resolution.
DeepGrid evaluated the decision stack across dark-store scenarios encompassing inbound put-away, pick runs, and dense peak-hour floor operations. The planner resolves obstacles including workers, peer automated guided vehicles, trolleys, and spills, using configurable graded alerts ranging from honks to silent operation. When routes are blocked, an explicit stuck-time threshold triggers automatic rerouting in simulation, not a fielded deployment.
- Simulation, not a fielded deploymentValidation status
- AMR Forklift Dark-Store SimDemonstrator platform
- Put-away, pick, peak floorSimulated scenarios
- Configurable honk, voice, silentAlert policy
- Auto-reroute on stuck-timeDeadlock recovery
Portfolio deck, slide 16: Proven in a simulatorPortfolio deck, slide 17: The simulator, running
What has to go right
The primary operational bottleneck for AD1 expansion is site mapping and commissioning throughput, which depends on field services capacity rather than pure silicon product availability. DeepGrid must also ensure that its drive-by-wire actuation interfaces seamlessly integrate across diverse third-party industrial vehicle models already owned by customer sites. Operationally, the software decision layer must reliably resolve dense real-world hazards, such as spills, peer automated guided vehicles, and workers, utilizing preset stuck-time auto-rerouting to prevent floor deadlocks. Scaling to 4,200 annual units by FY2032 also requires sustained distribution traction through third-party material-handling integrators.
Key dependency Site mapping and commissioning throughput, services capacity rather than product.
Further reading and viewing
Watch
The AD1 Indoor L4 Kit in the narrated walkthroughOne silicon. The full story. From 3:31Film: Indoor autonomyA warehouse truck that steers, brakes and stops for people · 1:12Read
The A100 ADAS SoC on the SoC2 dieTechnology: six domains, one tapeoutPortfolio deck, slide 100Where autonomy runsInvestment memorandum: BusinessAD1 Indoor autonomyInformation Memorandum v2 (Aug 2026), page 83 . T H E S O L U T I O NInformation Memorandum v2 (Aug 2026), page 134 . G O - T O - M A R K E T, P R I C I N GPortfolio deck, slide 6The portfolioInvestment memorandum: BusinessT100 AI licenceFinancial Model v3 (Sept 2026), workbookSheet: Revenue BuildAsk DeepGrid about the AD1 Indoor L4 KitAny question, answered from the same documentsRelated products
Prices, volumes, revenues and margins are management projections. The text is written from the linked documents, and every figure in it appears in the source it came from.