← All projects

Mechatronics · Automated tool changing

A peeler and an automatic blade-changing cutter.

I designed and built two food-preparation modules for a team kitchen-automation project: a peeler with independently driven blade angle and a cutter that exchanges blade discs automatically.

Lead Mechatronics Engineer | Sep 2024 - Jun 2026

My work

Mechanical design, control electronics, motion software, and prototype integration.

Demonstrated

Peeling runs on an apple and an orange, plus an automatic blade-change sequence.

Project stage

Built prototypes for a proposed modular preparation line.

Current peeling prototype | Physical hardware

Project brief

The team’s goal was a modular food-preparation line for commercial kitchens. My work focused on the first two machines: a peeler and an automatic blade-changing cutter. Washing and cooking were part of the broader platform concept.

The peeler rotates the ingredient while controlling blade travel and angle. The cutter uses a gantry to transfer blade discs from a rack to the cutting spindle. The built prototypes demonstrate peeling and automated blade exchange.

Program Modular commercial-kitchen automation
Mechanical scope Peeler with linear travel and two rotary motions; XYZ blade-changing gantry
Control scope ROS trajectory work, ESP32 motion control, and custom stepper electronics

4

Intended cutting modes

3-axis

Peeler, v2

ESP32 + Altium

Stepper-control PCB

ROS

Peeler v1 trajectory

Built, v2 Peeler

Driven fruit spindle, screw-driven linear travel, and an independently driven blade-angle axis.

Rebuilt and peeling Peeling motion

v1 lost consistent blade contact. v2 added blade-angle control and demonstrated peeling on an apple and an orange.

Built hardware Cutter redesign

The redesigned Cartesian cutter was modeled in Onshape and built with its blade-handling gantry, processor, and control hardware.

My role

As lead mechatronics engineer, I designed the mechanical architecture and PCB, implemented the controls, and built and integrated the cutting and peeling modules. I worked on the peeler and cutter within the broader team project.

System architecture

The proposed workflow runs from whole produce to peeling, then cutting. On the current peeler, a driven spindle rotates the fruit while linear blade travel and blade rotation control contact. The cutter exchanges discs to change the cutting tool.

The two modules were built as separate prototypes. Automatic ingredient transfer between them remains outside the demonstrated system.

Input Whole fruit
Physical prototype Peeling module
Built module Cutting module
Output Prepared ingredients

Peeler v1

The first peeling bench combined three motions: a lead-screw vertical axis, a rotating arm carrying the blade, and a spindle turning the fruit. The vertical axis used a 24 V stepper and the arm a Robodyno servo with position and velocity control.

I built the arm, then implemented inverse kinematics and helical-motion logic in ROS to coordinate vertical travel with rotation around the fruit.

I first checked motion against a printed spherical surrogate, then tried a real apple. The blade removed skin, but contact and cutting depth were inconsistent.

v1 contact trialArm sweeping a printed spherical surrogate before moving to real fruit

What v1 got wrong

The tests exposed positioning error and a blade-angle constraint.

The blade tip did not reach its commanded position consistently. I suspected gearbox backlash and flex in the 3D-printed motor adapter at the arm root. I did not measure their contributions separately, so I cannot assign the error to either one alone.

The blade also lacked independent angle adjustment. Its orientation was fixed relative to the arm, which made it difficult to maintain a suitable cutting angle as the fruit surface changed.

In the apple trials, the blade alternated between skimming the surface and digging into the flesh. The fixed blade orientation limited how much I could correct this through the commanded path.

I tried a pivoting blade with a rubber band to provide passive angle compliance. It did not maintain consistent contact in the trials.

Rigid blade against a real appleWith no degree of freedom at the blade, contact angle is set by the arm rather than by the fruit surface

Geometry and motion

I used the frame offsets and blade-tip reach in the assembly drawing to define the v1 arm geometry.

For the v1 motion model, I approximated the fruit as a sphere and described the rotating arm’s swept workspace as a cylinder. I used their intersection to relate vertical position to arm angle. This was a geometric approximation, not a measurement of each fruit’s surface.

The vertical and rotary axes needed coordinated speed profiles. The model supplied a nominal path, but could not correct gearbox play, adapter flex, or the fixed blade angle. I replaced the rotating-arm layout in v2.

Measured geometryRotating-arm assembly drawing and blade-tip offsets

Peeler v2

For v2, I replaced the rotating arm and added independent blade-angle control.

The fruit rotates on a motor-driven spindle whose position is fixed in the frame. A screw-driven carriage provides linear blade travel in place of the arm’s swept arc.

A separate motor controls blade angle. This allows me to adjust orientation independently of linear travel. The mechanism has one linear motion and two rotary motions, including the fruit spindle.

I replaced the Robodyno servo with a geared, open-loop stepper arrangement. The prototype ran without closed-loop position feedback.

The v1 rotating-arm inverse kinematics no longer applied. The revised sequence coordinates linear travel, fruit rotation, and blade angle.

The recorded v2 trials show skin removal on an apple and an orange. These demonstrate the mechanism on two examples; they do not establish batch throughput or a repeatability rate.

v2 benchExtrusion frame, supported lead screw, NEMA stepper and DM860H driver

Tool-changing cutter

Most food processors support several cutting styles, but an operator must stop the machine, open it, and replace the blade disc manually. That manual changeover interrupts a recipe that needs different cutting tools.

I designed and built the cutter around a Cartesian tool-changing mechanism. An XYZ gantry carries an electromagnetic gripper between the blade rack and cutter spindle. It lifts the selected disc, positions it above the spindle, and lowers it into place. A motorized lid clears the tool-change path and closes before cutting.

The cutter is designed to feed ingredients onto a dedicated 48 V spindle and release the cut product through an outlet door. Its intended cutting modes are slicing, dicing, chopping, and julienne, using different blade discs.

The built machine combines the 24V XYZ blade-changing gantry, electromagnetic pickup, indexed cutter spindle, motorized lid, outlet door, ESP32 control hardware, and separate 48V cutter drive in one frame.

The recorded demonstration shows the gantry selecting a disc, transferring it to the cutter, and seating it on the spindle. That demonstrates automatic blade handling; it does not by itself establish cutting performance across all four modes.

Redesigned cutterBuilt Cartesian cutting and blade-changing hardware
  1. 01 Select Choose a blade from the rack
  2. 02 Transfer Move the blade disc into alignment with the cutter shaft
  3. 03 Seat Lower the selected blade disc onto the cutter spindle
Automatic blade change, built hardwareFull sequence on the machine: selecting a disc from the rack, moving to the cutter, and seating it on the spindle

Control architecture

I designed an ESP32 stepper-control PCB in Altium. On peeler v1, I used ROS for the rotating-arm trajectory and inverse kinematics. Peeler v2 coordinates linear travel, fruit rotation, and blade angle.

The peeler firmware exposes serial commands for relative moves, homing, and peel sequences. The cutter uses ESP32-based control for the XYZ gantry, with TMC5160 stepper-driver channels and a separate cutter-spindle drive.

Peeler v1 ROS trajectory and inverse kinematics
Electronics ESP32 and custom Altium PCB
Cutter XYZ gantry and separate spindle drive
SolidWorks Onshape Altium ESP32 ROS Inverse kinematics Stepper motion
Open original image