Smart Truck (AD2 kit)
Smart Truck (AD2 kit) is a commercial-vehicle ADAS kit, forward camera, radar, compute box, driver display and vehicle harness, sold at ₹2,50,000 into the Indian N2/N3 truck market. It is the single largest SKU in the DeepGrid portfolio, representing a management projection of ₹450 Cr and 39.9% of group FY2032 revenue from one product line.
- Listed price
- ₹2,50,000
- FY2032 revenue
- ₹450 Cr
- Share of plan
- 39.9%
- Gross margin
- 89%
- First revenue
- FY2027
- Volume plan
- 18,000 @ FY32
From input to outcome
Proposed product architecture. The scene illustrates its application, not a shipped kit.
Camera and radar watch the forward path and the blind quarters
SoC2 fuses both into one object list with range and closing speed
Collision, lane and blind-spot logic decide whether to warn
Driver is warned, and the brake interface is signalled where fitted
Drag to rotate. Conceptual layout, not a mask.
On the die
The case in brief
- The ADAS mandate makes demand for the AD2 kit a compliance floor, not a forecast: 0.5 million new N2/N3 trucks per year plus 0.5 million retrofit units must be equipped, and DeepGrid's FY2032 target of 18,000 units is only 1.80% capture of that pool (management assumption).
- At a management-projected ₹450 Cr and 89% gross margin by FY2032, this single SKU represents 39.9% of group revenue, making ARAI or ICAT certification the decisive gate for the entire portfolio.
- The SoC2 compute box completes eleven-channel sensor fusion in 8.6 ms of a 33.3 ms frame, leaving 74% of the frame budget unused, the headroom that holds price flat at ₹2.50 L while sustaining 89% gross margin against imported alternatives.
What it is used for
Mandate compliance retrofit
- Buyer
- Fleet operators and retrofit installers
- The job
- Fleet operators running N2/N3 trucks face a compliance-led obligation under the ADAS mandate requiring driver-drowsiness, blind-spot and lane-departure capabilities on their vehicles; the decision is not discretionary.
- What DeepGrid supplies
- DeepGrid supplies the complete AD2 kit, camera, radar, compute and harness, through fleet-direct and retrofit installer channels, enabling fitment without waiting for OEM line-fit programmes.
OEM line-fit for new N2/N3 trucks
- Buyer
- Truck OEMs and OEM homologation engineers
- The job
- Truck OEMs must equip every new N2/N3 vehicle with certified ADAS to satisfy the mandate, requiring a kit that has cleared ARAI or ICAT certification and can be integrated at the production line.
- What DeepGrid supplies
- DeepGrid offers the same AD2 architecture certified to AIS-162/188, sold to OEM or Tier-1 at the listed price, with the compute module designed for a direct swap to owned 28 nm silicon without requiring a vendor change.
Driver drowsiness and attention compliance
- Buyer
- Fleet safety manager, OEM homologation engineer and testing-agency assessor
- The job
- Under AIS-184 (DDAWS) and CMVR 125Q(1), manufacturers must conduct a ten-participant day-and-night study generating ten or more true-positive or false-negative events and report per-participant sensitivity, approval is not a track test.
- What DeepGrid supplies
- DeepGrid supplies an in-cab drowsiness pipeline covering eyelid closure, gaze, head pose and PERCLOS-class metrics fused with vehicle signals, together with the completed documentation package, metrics, thresholds, activation speed, weather limits and statistical analysis, that makes the submission approvable.
The story
The AD2 kit is a complete commercial-vehicle ADAS system sold at ₹2.50 L into the Indian N2/N3 truck market.
Smart Truck is a bolt-on ADAS kit comprising a forward camera, 4D radar, SoC2 compute box, driver interface display and vehicle bus harness, everything needed to bring a commercial truck into compliance in a single fitment. At ₹2.50 L it is the largest SKU in the DeepGrid portfolio, with first revenue targeted in FY2027. Management describes it as the reason the portfolio carries a demand floor rather than a demand forecast.
- ₹2.50 Lunit price
- Systemssegment
- Outdooroperating domain
- FY2027first revenue
- Forward camera + 4D radarsensor components
- SoC2 compute boxcompute component
Camera and radar feed a single fused object list in 8.6 ms, giving the decision layer range and closing speed before a braking window closes.
The forward camera and 4D radar monitor the truck's forward path and blind quarters simultaneously; the SoC2 box fuses both streams into one object list carrying range and closing speed for every detected target, completing that work in 8.6 ms of a 33.3 ms frame. Collision, lane-departure and blind-spot logic then decide whether to warn, alerting the driver visually and acoustically and signalling the brake interface where the vehicle supports it. The 74% of frame budget left unused is what allows one hardware part to serve every sensor configuration below it in the product family.
- 11 inputs simultaneouslycompute channels
- 7 × 8MP RGB, GMSL2camera inputs
- 2 thermal camerasthermal inputs
- 8.6 ms fusion latencysensor fusion time
- 74% frame budget unusedcompute headroom
- ARAI or ICATcertification gate
Fleet operators and truck OEMs must buy, the ADAS mandate makes this a compliance obligation, not a discretionary purchase.
The ADAS mandate imposes driver-drowsiness, blind-spot and lane-departure obligations on N2/N3 commercial vehicles, so the buying decision for fleet operators and truck OEMs is which supplier to use and on what date, not whether to act. DeepGrid reaches fleets directly and through retrofit installers now, with OEM line-fit as the higher-volume route over time. The combined demand pool, 0.5 million new N2/N3 trucks per year plus 0.5 million retrofit units, is drawn from third-party published estimates; the FY2032 volume target of 18,000 units represents a management assumption of 1.80% capture of that pool.
- Fleet operators + truck OEMsbuyer
- Compliance-ledpurchase trigger
- Fleet direct + retrofit installersnear-term channel
- OEM line-fitlonger-term channel
- 0.5M new N2/N3 trucks/yrnew-vehicle pool
- 0.5M retrofit unitsretrofit pool
One SKU accounts for 39.9% of group FY2032 revenue at 89% gross margin, the concentration is the portfolio's largest single bet.
At a management projection of ₹450 Cr by FY2032, the AD2 kit is the largest line in the plan and the anchor of the group's demand floor. The 89% gross margin is a management projection, and the unit price is held flat across all plan years, so any margin expansion comes from replacing FPGA-era bill-of-materials costs with the owned 28 nm SoC2 die. Certification through ARAI or ICAT is the gate that converts regulatory demand into bookable orders; without it the OEM line-fit channel, which is the volume engine, cannot open.
- 18,000 unitsFY2032 units (mgmt proj.)
- ₹450 CrFY2032 revenue (mgmt proj.)
- 39.9% of group revenueportfolio share
- 89% gross marginmargin (mgmt proj.)
- ARAI or ICAT cert.primary volume gate
- ₹0.07 CrFY273 units
- ₹2 CrFY2880 units
- ₹35 CrFY291,400 units
- ₹100 CrFY304,000 units
- ₹225 CrFY319,000 units
- ₹450 CrFY3218,000 units
The production perception stack runs against three road profiles in a live dynamics simulator, simulation, not a fielded deployment.
D-DRIVE · ADAS Sim v2 runs the same perception stack across a market street, a single-carriageway highway and a deliberate stress-test scenario without reloading between them; the bird's-eye planner view redraws continuously from the fused object list, not from a scripted animation path. The vehicle plant model uses a 550 kg mass, 2.10 m wheelbase and tyre friction of μ 0.75, with steering angle and lateral acceleration solved rather than scripted. The perception rail names detected classes with live range, a motorbike and an auto-rickshaw, metres out, showing the decision layer a buyer is actually purchasing before any truck is fitted.
- D-DRIVE · ADAS Sim v2demonstrator
- 3 road profiles, 1 stackscenario coverage
- 550 kg massplant model mass
- 2.10 m wheelbaseplant model wheelbase
- μ 0.75 tyre frictionplant model friction
- Simulation, not fielded deploymentevidence status
Portfolio deck, slide 22: Proven in a simulatorPortfolio deck, slide 23: The simulator, running
What has to go right
Certification through ARAI or ICAT is the primary gate: without it the OEM line-fit channel, which is the volume engine, cannot open, and a delay also affects the AD0 Smart Mirror family sold to overlapping buyers, together those two SKUs represent ₹720 Cr of projected FY2032 revenue, more than the other thirteen SKUs combined. Installation channel capacity must scale to 18,000 fitments per year, a logistics and installer-network build that has no precedent in the current business. The ten-participant day-and-night drowsiness study required for AIS-184 approval must be completed and documented to a standard that a testing-agency assessor can act on, not merely demonstrated. ASP is held flat in the model, so any price erosion from competitive or channel pressure flows directly to margin with no volume offset built in.
Key dependency AIS-mandate compliance, installation channel capacity, and radar verification milestones.
Further reading and viewing
Watch
The Smart Truck (AD2 kit) in the narrated walkthroughOne silicon. The full story. From 5:25Film: The truck kit’s sensorsMirror towers, cab cameras and 4D radar on a heavy truck · 0:46Film: Eleven sensors, one outputThe 8.6 ms frame budget on the compute box · 0:54Film: D-Drive on three road profilesVehicle physics solved live, not a recorded drive · 1:18Read
The A100 ADAS SoC on the SoC2 dieTechnology: six domains, one tapeoutHow eleven sensors fit an 8.6 ms budgetTechnology: sensor to computeCommercial use case UC-04Driver drowsiness and attentionPortfolio deck, slide 98ConcentrationPortfolio deck, slide 6The portfolioPortfolio deck, slide 24Eleven inputsBP1A India Autonomous Trucking Plan, page 8Certified AD2 line-fit kit (the mandate product): same architecture, AIS-162/188 certifiedBP1A India Autonomous Trucking Plan, page 27What sits inside the ADAS numberAsk DeepGrid about the Smart Truck (AD2 kit)Any 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.