The Build

How We Built SPARSH

SPARSH is a Cartesian CNC drawing robot, built from standard parts available in any Indian electronics market. Here is what it is made of, how it is programmed, and the problems we solved to make it reliable.

What It's Made Of

Bill of Materials

ComponentPurposeEst. (₹)
Arduino Board (Brain)Main controller for all robot operations500
NEMA 17 Stepper MotorsPrecise, repeatable movement for accurate drawing1,200
DRV8825 Stepper Drivers32-step microstepping for smoother lines150
MG996R / SG90 ServosPen lift and automatic tool changer450
XYZ Joystick ModuleDirect artist input — Co-Creation Mode120
LEGO Technic / Acrylic / MDF FrameStructure — strength, adjustability, stability2,000
Lead Screw & NutConverts motor rotation to precise linear motion, no backlash700
CNC Shield V3Connects Arduino to drivers — reduces wiring errors250
Limit SwitchesHome position and overtravel protection100
X-Axis & Y-Axis Gantry AssemblyCartesian motion — reaches any point on A2 paper2,500

A complete SPARSH system costs approximately ₹15,000–₹20,000 to build, including the tool changer, 3D-printed parts, and consumables. A single large commissioned artwork can sell for ₹15,000–₹20,000 — so one painting can recover the entire cost of the system.

The Software

SPARSH Engine & Dashboard

The SPARSH dashboard was built in PictoBlox and Python — written entirely by Nysa and Neev. It is the control screen where the artist selects a mode, chooses artwork, monitors the robot, and manages pen and tool changes.

Mode Selection

Triggers the joystick processing loop or the voice recognition listener

Artwork Library

Stores outline coordinate files matched to voice commands via a library index

Voice Command Input

PictoBlox captures the spoken word, matches the library index, sends the design ID

Robot Status

Displays Ready, Drawing, Tool Change, Paused, or Completed

Tool / Colour Status

Shows which tool slot is currently locked

Emergency Stop

Overrides all motor outputs immediately

AI Declaration — WRO Rule 6.5

PictoBlox speech recognition matches spoken words to a fixed design library only — no generative AI is used anywhere in SPARSH. All motion control uses mathematical calculation. Every line of code was written by Nysa and Neev.

What We Tested

Execution Results

Voice Recognition

SPARSH recognised the command "Draw Sunflower" and selected the correct design from the library.

Co-Creation Mode

Artist-controlled joystick movements were accurately converted into CNC drawing motion.

AI-Assisted Mode

SPARSH automatically generated and drew the selected sunflower outline on paper.

Automatic Tool Changer

The robot picked, locked, and switched between different coloured pens without assistance.

X-Y-Z Motion Control

Coordinated X, Y, and Z motion held accuracy across the full drawing area.

Large-Format Artwork

SPARSH produced A2-size artwork while maintaining drawing accuracy throughout.

What We Solved

Challenges We Faced

Voice Recognition in Noise

In noisy rooms the robot misheard commands and picked the wrong shape. We tested across noise conditions, adjusted sensitivity, and trained the recogniser on multiple voices for more reliable matching.

Drawing Accuracy Drift

Over time the movement system loosened and lines drifted from where they should be. We tightened the V-slot connections and drive regularly, and added a calibration check at the start of every session.

Pen Pressure

Different markers press differently — some tore the paper, some drew too faint. We tuned the servo angle limit so the pen meets the paper with the right pressure, and tested across marker types.

Smooth Drawing

Sharp direction changes made corners look rough and jagged. We improved path smoothing in the SPARSH Engine so the robot eases slightly before a hard line change.

Multi-Pen Operation

The tool changer sometimes grabbed the wrong marker or missed the slot. We calibrated each pickup position precisely and added an automatic retry if the first attempt fails.

Built by Two Grade 5 Students

Every part of SPARSH was designed, assembled, coded, and tested by Nysa and Neev — guided by our mentor and shaped by real artists at every step.

Meet the Team