Edge Capstone — Research Robot

What you'll build

A complete system design for a swarm of low-cost ground robots that monitor an Indian smallholder farm — soil moisture, pest activity, crop health — and produce a daily report for the farmer. You'll deliver: a 3-page architecture document, a ROS2 implementation sketch, deployment considerations specific to rural India, and an honest risks-and-tradeoffs analysis.

This is the lesson where there's no answer key. You're designing, not following. The artefact is the deliverable — and a strong version of it is what a research PI at IIT Bombay or a senior engineer at Niqo Robotics would expect from a PhD applicant.

The problem

Indian agriculture is dominated by smallholders — average farm size is 1.08 hectares (vs ~50 hectares in the US). Existing precision-agriculture tools (John Deere monitors, Niqo's tractor-mounted sprayers) assume scale and capital that smallholders don't have. The market for sub-₹50,000 monitoring solutions is open — and barely served.

A swarm of 5–10 small ground robots (~₹15k each) that together cover a 1-hectare field daily could give a farmer:

• Soil moisture maps (drive sensors into the ground at sample points)
• Pest scout images (cheap cameras flag visible damage)
• Crop-height differential (where is growth lagging?)
• A morning WhatsApp summary

It's a real research question. TiHAN at IIT Hyderabad is funding work in this space. Niqo Robotics is interested in the ground-truth data. ICAR-CIAE (Central Institute of Agricultural Engineering) in Bhopal would partner.

What to deliver

Section A: Problem framing (~500 words)

• Who is the user? (smallholder, district co-op, agritech startup, govt research)
• What do they pay for today, and what would they pay for this?
• What does "good enough" look like in v1?

Section B: System architecture (1–2 pages, sketches OK)

• Individual robot — sensors, compute, motors, power, battery, cost target
• Fleet — how do they coordinate? Centralised farmer's-phone hub or peer-to-peer?
• Communication — 4G, LoRaWAN, or wifi from a portable router?
• Data pipeline — on-robot processing vs cloud sync vs local edge gateway

Section C: ROS2 implementation sketch (~1 page)

• Top-level node graph (which nodes run on robot vs on hub)
• Topics, services, actions — what flows between them
• ROS2 vs micro-ROS choice (likely a mix — micro-ROS on the robot, ROS2 on the hub)
• Multi-robot namespacing scheme (revisit Edge 01)

Section D: Deployment realities (~400 words)

• Rural connectivity — what works at 4G/LTE-NB-IoT/ no signal?
• Power — solar charging stations, swappable batteries, both?
• Weather — monsoon, dust, 45°C heat
• Failure modes — what happens when one robot loses GPS or rolls into a ditch?
• Support — who fixes a broken robot in a village 200km from Bangalore?

Section E: Risks & honest tradeoffs (~300 words)

• Why might this fail commercially even if the tech works?
• What's the biggest technical risk?
• Three things you'd test in a 30-day pilot before scaling

A starter ROS2 sketch

Per robot:

┌─ Sensors ─┐   ┌─ Perception ─┐   ┌─ Behavior ─┐   ┌─ Comms ─┐
│ /camera   │──▶│ /detections  │──▶│ /next_goal │──▶│ /upload │
│ /soil     │──▶│ /soil_signal │──▶│            │   └─────────┘
│ /gps      │──▶│ /pose        │──▶│            │
└───────────┘   └──────────────┘   └────────────┘
                                          │
                                          ▼
                                       Nav2 (cut-down)

Hub (the farmer's Android phone + cheap raspberry-pi gateway):

┌─ /upload (all robots) ─┐   ┌─ /reports ─┐   ┌─ WhatsApp ─┐
│                        │──▶│ aggregate  │──▶│ send daily │
└────────────────────────┘   └────────────┘   └────────────┘
            │
            ▼
        Plot map / dashboard

Communication: LoRaWAN (free, long range, low bandwidth — perfect for soil readings + GPS + status) for telemetry, then sync image data over wifi when robots return to a charging station.

Why this matters in 2026

In 2025, India launched the National Mission on Agricultural Robotics under the Ministry of Agriculture, allocating ₹1,200 Cr for precision-ag R&D over 5 years (announced Budget 2025). Universities and startups are bidding for grants. A research-quality system design for smallholder swarms is fundable work right now.

If your design is rigorous, the next step is talking to:

• TiHAN-IIT Hyderabad (autonomy fellowships)
• ICAR-CIAE Bhopal (agri partnerships)
• Niqo Robotics, EFKON, Tractors and Farm Equipment Ltd (TAFE) — companies that want to license tech

Test Your Understanding

This time it's not 3 questions — it's the design document. Below are the only three rubric items:

1. A technical reviewer (PI at IIT Madras) reads your design and says: "I see how it works. But why won't this fail in the monsoon?" — How do you respond? Be specific about ingress protection, electronics, and fallback modes.

2. A business mentor asks: "Show me three measurable outcomes the farmer will care about after one season." — Each outcome should be (a) numerically defined, (b) measurable by the system, (c) tied to a rupee benefit the farmer can describe to their family.

3. A senior robotics engineer reads your ROS2 sketch and asks: "Why does each robot run Nav2? That's a lot of compute for a ₹15k robot." — Defend your choice or propose an alternative. There's no single right answer — be specific.

India Opportunity

If you deliver a serious version of this capstone, you're hireable / fundable for:

• PhD candidate (Agri-Robotics) · IIT Bombay / IIT Madras / IIT Kharagpur — full stipend, 5 years.
• Research Engineer · TiHAN-IIT Hyderabad — fellowship-funded research projects, ₹70k–1L/month.
• Founding Engineer · agritech startup pre-seed (YC India, Antler, Sequoia Surge) — equity + ₹15–30 LPA.
• Senior Research Engineer · Niqo Robotics, Bharat Electronics — applied ag-robotics research, ₹24–40 LPA.

You finished the Edge track

There's no more lesson after this one. Add the design document to your portfolio. Submit it to the /showcase page on R2BOT — we feature strong work. Apply for research positions and grants. Email it to PIs at IITs.

Robotics in India is being built right now. Your name should be on it.

→ Or revisit: Edge 01 · Multi-Robot Systems · Edge 02 · SLAM from Scratch.