Embedded Files
CAD Design
CROPMINDER MARK II
CROPMINDER MARK II
The project develops a compact autonomous Unmanned Ground Vehicle (UGV) that integrates fruit detection, depth estimation, inverse kinematics, and actuation into a single deployable harvesting system.
A YOLOv8n deep learning model is trained to detect ripe strawberries and classify leaf diseases, enabling selective harvesting and basic crop health monitoring in real farm conditions.
Using a calibrated pinhole camera model, detected strawberries are converted from image coordinates into real-world 3D positions, enabling accurate fruit localization without expensive LiDAR sensors.
A Jacobian-based inverse kinematics controller computes joint angles for dual 6-DOF robotic arms, translating spatial targets into precise servo-level motion for controlled picking.
The system runs on a Raspberry Pi–Arduino architecture with dedicated motor drivers and servo controllers, creating a scalable, cost-effective automation solution tailored for small-scale strawberry farms.
The Project was recognized as the BEST HARDWARE IMPLEMENTATION at Techovate 25 conducted by Amity University
DEVELOPMENT COST = $ 2400
DEVELOPMENT DURATION = JUN 2024 - MAY 2025
Custom CNN for Ripeness Identification
Dimensions
Width - 350mm, (600mm in full stretch)
Height - 500mm (800mm in full stretch)
Length - 950mm
Specifications
Weight - 30kg
Max Speed - 10kmph
Battery Weight - 5kg
Runtime - 3 hrs (single battery)
Extended Runtime - 3 hr 20 mins (with regen)
Total Recommended Runtime - 10 hrs (3 Ideal battery swaps)
Basket Load - 5 kg (single load)
The concept of affordable automation is a driving factor in building this agri-tech product
Inverse Kinematics Programming
CAD Drawing
Cropminder Mark I
Electronics of Mark II
Depth Estimation via Two Inputs
MEDIA COVERAGE
AGRIPULSE - Forward Deployed AI Analyst for Ag-Businesses
AGRIPULSE - Forward Deployed AI Analyst for Ag-Businesses
AgriPulse is a U.S. agricultural intelligence platform that combines commodity prices, crop production, weather, trade, and farm decision support in a single interface.
Developed an interactive workflow where users select a state, crop, acreage, and growth stage to view production statistics, 20-year price trends, weather risks, and AI-generated recommendations.
Frontend uses React, TypeScript, Vite, Recharts, React Simple Maps, Leaflet, React Router, and Lucide React, and deployed it on Vercel.
Backend uses Node.js, Express, TypeScript, Prisma, and Supabase PostgreSQL, with a separate Python FastAPI intelligence service.
Database integrates USDA NASS, USDA FAS GATS, and Open-Meteo data, while the Farmer Copilot combines regional, production, pricing, and weather context to return recommendations, alternatives, sources, limitations, and confidence scores.
Explore it here:
2026-08-01 17-42-08.mp4Demo Video
DOSE AWARE RRT* PATH PLANNING ALGORITHM FOR UUV INSIDE NUCLEAR SPENT FUEL AND REACTOR POOL
DOSE AWARE RRT* PATH PLANNING ALGORITHM FOR UUV INSIDE NUCLEAR SPENT FUEL AND REACTOR POOL
Built a path planning algorithm based on RRT* for autonomous UUV that can safely inspect nuclear reactor or spent-fuel pool environments while avoiding dangerous radiation and heat.
Used OpenMC simulations to generate realistic radiation and temperature maps of the reactor environment.
Converted simulation data into a 3D point cloud in ROS2/RViz to create a virtual nuclear inspection environment.
Simulated a URDF-based robot in ROS2 and track its movement within the mapped radiation field.
Applied a dose- and temperature-aware RRT* algorithm to plan safe, optimal paths for autonomous inspection.
Further information available in Research Paper under review at IEEE Transactions of Nuclear Submission
Github link below
Digital Twin CAD of Fuel Rod Assembly at both Reactor Core and Spent Fuel Pool
Dose Voxels Inside the Pool completing the digital twin
Extracted Radiation and Temperature Dose via Openmc
Radiation and Temperature voxels after threshold setting
UUV Path Planning using Odometry and RRT*
OMNI-DIRECTIONAL AGV FOR WAREHOUSES AND INDUSTRIAL APPLICATIONS
OMNI-DIRECTIONAL AGV FOR WAREHOUSES AND INDUSTRIAL APPLICATIONS
Conceptual design for a fully autonomous logistics AGV inspired by forklift operations, featuring mecanum-wheel omnidirectional mobility, adjustable air-suspension chassis for load engagement, dual LiDAR-based perception, and full 360° safety coverage for industrial warehouse environments.
This is my own design replica of "Fraunhofer O3dyn" and is used solely to showcase my skills in design, control systems, and embedded systems. I do not intend to commercialise or open-source this design. Design and patent copyrights exist with the actual owner.
A SATELLITE-BASED SYSTEM FOR SOLAR FARM SITE ASSESSMENT
A SATELLITE-BASED SYSTEM FOR SOLAR FARM SITE ASSESSMENT
The Project is a satellite-driven framework that evaluates how suitable a location is for large-scale solar power installation using open-source Earth observation and climate data.
It combines key environmental factors such as solar irradiance, terrain slope and elevation, vegetation density (NDVI), and soil moisture extracted for specific geographic coordinates.
Each parameter is normalized and processed through a physics-informed model to generate a clear Solar Suitability Index (SSI) from 0–100 for quick feasibility assessment.
The system converts raw satellite data into transparent, scalable, and actionable insights to support early-stage solar farm planning and site selection.
FURTHER INFO IN RESEARCH PAPER CURRENTLY UNDER REVIEW AT IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
Github link below
Bhadla Solar Park, Rajasthan, India
Hudspeth County, Texas, USA
Thurston County, Washington, USA
The framework produced suitability scores consistent with real-world solar deployment patterns.
High-irradiance desert regions achieved scores above 90, indicating strong feasibility for large-scale photovoltaic installation, while densely vegetated, Low-irradiance regions scored significantly lower. These results validate the model’s ability to reliably distinguish optimal and suboptimal solar development locations using satellite-derived environmental data.
DIGITAL TWIN FOR ULTRA FAST CHARGING OF 625AH LITHIUM ION BATTERY
DIGITAL TWIN FOR ULTRA FAST CHARGING OF 625AH LITHIUM ION BATTERY
This Simulink model represents a digital-twin-based lithium-ion battery charging and monitoring architecture with anode-potential safety enforcement. Measured pack voltage and current are injected with configurable disturbances and fed into a twin estimator that reconstructs internal states such as SOC, anode potential, overpotential, RC voltage, and adaptive ohmic resistance.
A parameter-adaptation block updates model parameters online and computes residual errors used for fault detection. Constraint monitors translate estimated electrochemical limits into current bounds, which gate the CC–CV current request.
The architecture explicitly separates estimation, adaptation, constraint enforcement, and fault flagging to ensure robust ultra-fast charging under sensor noise and model mismatch.
SUSPENSION INTEGRATED BELT DRIVEN TRACK SYSTEM FOR UGV'S
SUSPENSION INTEGRATED BELT DRIVEN TRACK SYSTEM FOR UGV'S
Designed a compact belt-driven tracked propulsion system with integrated suspension and triple coil-spring bogies for improved stability and traction on uneven terrain. Intended as a modular mobility platform for agricultural and reconnaissance UGVs, optimized for low-profile integration and harsh outdoor conditions. Concept validated through virtual modeling only (pre-fabrication stage).
Tech stack: Fusion 360 (3D CAD), kinematic simulations (suspension + track path), CAM-oriented design for CNC or laser-cut components; prepared for future integration with motor drivers, encoders, and embedded controllers.
DYNAMIC POWER LOSS MODELLING IN TRACKED VEHICLE SUSPENSION (MATLAB PROJECT)
DYNAMIC POWER LOSS MODELLING IN TRACKED VEHICLE SUSPENSION (MATLAB PROJECT)
Designed and simulated a control system in MATLAB to model power transmission losses in a tracked vehicle under dynamic suspension compression. Integrated PID control logic to maintain target power output, accounting for mechanical inefficiencies and terrain-induced suspension variations. Visualized system efficiency, compression behavior, and control responses through multi-plot analysis.
ZAFFRA - SAFFRON HARVESTER MACHINE FOR PRECISION AGRICULTURE
ZAFFRA - SAFFRON HARVESTER MACHINE FOR PRECISION AGRICULTURE
Designed ZAFFRA, a multi-arm robotic system concept for automated saffron harvesting using four 6-DOF arms: dual stigma-harvest arms, a bulb-stabilizing soft gripper arm, and a petal-removal arm synchronized with a conveyor for continuous processing.
Proposed vision and control: NVIDIA Jetson Orin Nano, Arducam Sony IMX477 cameras, multi-model saffron part detection, depth estimation, inverse kinematics for coordinated arm motion, shadowless lighting for stable imaging, and an integrated 7" HMI for calibration and system tuning.
Target performance: precise isolation of stigma, petals, and bulbs within a 25 cm workspace per arm, reduced manual labor for high-volume saffron farms, consistent extraction quality under controlled lighting, and optimized power envelope of 12V 30A for deployable, production-ready units.
The product is currently in the R&D stage.
Designed a compact indoor hydroponic system concept for apartments and restaurants to enable year-round, pesticide-free vegetable production with automated control of nutrients, lighting, and climate in a minimal footprint.
Tech: STM32 microcontroller (embedded C++), custom AC–DC regulation & protection circuitry, pH & TDS sensing, temperature & humidity monitoring, water pumps & aeration, fans & heaters, full-spectrum LED grow lighting, 15" touchscreen HMI, wireless data streaming for remote monitoring.
Expected Impact: Provide fully closed-loop environmental and nutrient control on standard 110–240V AC, targeting >90% reduction in manual intervention, stable crop yields independent of season, and reliable, on-demand local production for small spaces.
HYDROPOD - A MODERN HYRDROPONIC POD SYSTEM FOR URBAN APARTMENTS
HYDROPOD - A MODERN HYRDROPONIC POD SYSTEM FOR URBAN APARTMENTS
Above shown is the Rendered Image of the Concept
Designed and Developed a Hexacopter with a Day and Night Surveillance system with an inbuilt face detection system, which could be used for riot control and border surveillance. this drone includes 500g of weight carrying capacity making it able to carry pocket-size grenades and other equipment. the 40-minute flight time helps to maintain a steady flight for long-duration rescue missions.
DEVELOPMENT COST = $ 510
DEVELOPMENT DURATION = APR 2023 - AUG 2023
DAY AND NIGHT SURVILLENCE DRONE WITH FACIAL RECOGNITION
DAY AND NIGHT SURVILLENCE DRONE WITH FACIAL RECOGNITION
FPV (FIRST PERSON VIEW) DRONE BUILD
FPV (FIRST PERSON VIEW) DRONE BUILD
DEVELOPMENT COST = $ 300
DEVELOPMENT DURATION = AUG 2023 - AUG 2024
IMG_6938.mov
DESIGN OF SMART BULLETPROOF VEST WITH REAL-TIME HEALTH MONITORING AND CASUALTY ASSESSMENT (FEA)
DESIGN OF SMART BULLETPROOF VEST WITH REAL-TIME HEALTH MONITORING AND CASUALTY ASSESSMENT (FEA)
Integrating Force Sensors, Accelerometers, and Pulse Sensor to an ARM STM32F103C6 to transmit the real-time data of force being applied to suit and to measure the health of the Military Personnel.
Deep - Tech Explanainers
Deep - Tech Explanainers
Page updated
Google Sites
Report abuse