NVIDIA Jetson Nano
- Specification: Embedded GPU computer (Ubuntu 18.04 + PyTorch)
- Purpose: Runs CNN weed detection, logs soil data, sends movement commands.
System Introduction
System introduction, objectives, and architecture overview of the Smart Agri Four Legged Bot.
Project Abstract
Section titled “Project Abstract”The Smart Agri Four Legged Bot is a semi-autonomous agricultural robot designed to support precision farming through automated weed detection and in-situ soil condition monitoring. The project emphasizes modularity, mechanical reliability, and ease of maintenance under real-world agricultural constraints.
The robot is engineered around a hybrid fabrication approach:
-
Mechanical Substructure
A robust, laser-cut 3mm mild steel chassis reinforced with bottom plates and connected to four articulated planar legs.
-
Electronics & Computing
An NVIDIA Jetson Nano single-board computer acts as the central brain, handling visual input from an Intel RealSense D435 depth camera and serial telemetry from an Amici Sense soil probe.
-
Actuation & Low-Level Control
An Arduino Due manages low-level motor control, driving four planetary geared DC motors for shoulder articulation and two distal brushless hub motors for locomotion.
Project Objectives
Section titled “Project Objectives”- Terrain Adaptability: Design a mechanically stable, terrain-capable chassis optimized for uneven, sloped agricultural fields.
- Structural Verification: Validate the mechanical load paths, stress limits, and deflections using Finite Element Analysis (FEA) to guarantee structural safety.
- Hybrid Fabrication: Construct load-bearing structures with steel while using FDM 3D printing (PETG/PLA+) for lightweight, non-structural covers and sensor enclosures.
- Vision-Based Weed Detection: Integrate a real-time ML pipeline using deep learning (lightweight CNN) for weed bounding box classification on the edge.
- Precision Soil Probe: Integrate an in-ground multi-parameter sensor to measure soil health parameters without tipping or stability issues.
- Resilient Electrical System: Implement EMI signal isolation, separate power rails, and robust wiring layouts to ensure continuous field operation.
System Architecture
Section titled “System Architecture”The following block diagram illustrates the information flow, power rails, and control interfaces of the robot:
graph TD
%% Power Section
subgraph Power System
Battery["24V 12Ah Li-ion Battery"] --> Switch["Master Kill Switch"]
Switch --> Buck5V["DC-DC Buck Converter 5V"]
Switch --> Reg12V["DC-DC Regulator 12V"]
Switch --> Direct24V["Direct 24V Power Bus"]
end
%% Compute Section
subgraph Compute & Logic
Jetson["NVIDIA Jetson Nano"]
Arduino["Arduino Due"]
Jetson <-->|UART 115200 Baud| Arduino
end
%% Perception Section
subgraph Sensors & Input
Camera["Intel RealSense D435 Depth Camera"] -->|USB 3.0| Jetson
Probe["Amici Sense Soil Probe"] -->|UART GPIO| Jetson
end
%% Actuation Section
subgraph Actuation & Drive
Drivers["Dual H-Bridge Motor Drivers"]
ESCs["High-Current ESCs"]
Arduino -->|PWM Signals| Drivers
Arduino -->|PWM Signals| ESCs
Drivers -->|12V Rail| GearedMotors["4x Geared Shoulder Motors"]
ESCs -->|24V Rail| HubMotors["2x Distal Locomotion Hub Motors"]
end
%% Power Routing
Buck5V -->|5V Power| Jetson
Buck5V -->|5V Power| Camera
Buck5V -->|5V Power| Probe
Reg12V -->|12V Power| Drivers
Direct24V -->|24V Power| ESCs
Internal Layout CAD Detail
Section titled “Internal Layout CAD Detail”The internal layout was designed to separate data processing units from high-current power electronics to minimize EMI and facilitate thermal convection.
Kinematics & Leg Articulation
Section titled “Kinematics & Leg Articulation”The robot’s planar legs are articulated in three segments, allowing the body to adjust its pitch and ground height to maintain vertical camera stability and traction over sloped terrain.
Leg Mechanical Linkages & Compliance
Section titled “Leg Mechanical Linkages & Compliance”Each planar leg is engineered as a modular three-joint assembly (Hip, Knee, and Foot) with passive compliance and internal routing pathways:
- Torsional Spring compliance: A torsional spring joint is embedded in the knee joint housing to act as a passive compliance mechanism, helping to absorb ground vibration and keep traction over uneven profiles.
- Split-Piece Routing Layout: Thigh and shin segments feature a split-piece layout that allows routing high-current lines and encoder cables internally, protecting wires from exposure and mechanical wear.
Key Hardware Summary
Section titled “Key Hardware Summary”Arduino Due
- Specification: 32-bit ARM Cortex-M3 microcontroller (84 MHz)
- Purpose: Performs real-time low-level motor actuation and sync.
Intel RealSense D435
- Specification: RGB-D Active Infrared Stereo depth camera
- Purpose: Captures depth profiling and RGB frames for ML inference.
Amici Sense Probe
- Specification: Dual 180mm stainless steel probes
- Purpose: Measures soil moisture, pH, fertility, temperature, light, and humidity.
Geared DC Motors
- Specification: 4x planetary geared units
- Purpose: Drives shoulder joints for body height and pitch adjustments.
Brushless Hub Motors
- Specification: 2x 36V 350W direct-drive wheels
- Purpose: Distal foot units providing primary forward and reverse locomotion.
Project Photo Gallery
Section titled “Project Photo Gallery”Below are selected high-resolution photographs taken during the assembly and field testing phase of the physical prototype:
