Annin Robotics Community

At this stage of the PaMiLoCELL project, I am carrying out a high-level brainstorming exercise to understand how the robotic cell could evolve before freezing the final layout.

The objective is not only to define the physical arrangement of the cell, but also to evaluate whether the project can develop into a technically credible and potentially fundable concept.

Step 0 – AR4 Operating Envelope Analysis

The starting point is the AR4-MK5 working envelope: maximum reach, inner non-reachable area, joint limits and the difference between theoretical reach and the actually useful workspace.

Step 1 – Base Cell Layout

The first concept uses a central inspection/pick area with sorting bins distributed around the perimeter, keeping the first development phase simple and easy to validate.

Step 3 – Safety + AI Vision Integration

The next evolution introduces the enclosure, E-stop/reset functions, PLC/safety architecture and AI Vision for object recognition and material sorting.

Step 4 – Modular Expansion with Conveyor

A modular/openable side could allow a small conveyor to be added later without redesigning the entire cell, moving from a static demonstrator toward a more dynamic sorting system.

Beyond the layout itself, I am also starting to look at the project from an innovation and funding perspective:

• What is genuinely innovative compared with the current state of the art?

• Which technical problem does the solution solve, and for whom?

• Can the concept improve existing solutions in performance, flexibility, safety, maintainability or cost?

• Is there a real market/user need behind the technical idea?

• How does it compare with existing and competing solutions?

• What are the main technical and market risks?

• Which elements could generate proprietary know-how, IP or potentially patentable solutions?

• Can the concept be validated through measurable technical and economic KPIs?

This changes the brainstorming process: the goal is no longer only “Can I build it?”, but also “Why should this solution exist, and what measurable value can it create?”

Questions I am currently asking myself:

1. What innovative aspects can I introduce compared with the current state of the art?

2. Is reducing construction and maintenance costs alone enough to create meaningful innovation?

3. Which measurable benefit would make this solution attractive to a real industrial user or investor?

4. Which technical result should I demonstrate first to reduce the project risk and strengthen a future funding application?

Note: The images shown in this post are AI-generated concept simulations created to support layout studies.

Analyze → Design → Validate → Improve

#PaMiLoCELL #PaMiLoLAB #Robotics #IndustrialAutomation #RobotCell #AR4MK5 #AnninRobotics #AIVision #MachineVision #FunctionalSafety #PLC #AutomationEngineering #MaterialSorting #Innovation #TechnologyTransfer #OpenSourceRobotics

While assembling the AR4-MK5 and studying industrial robots, I am realizing that robot motion is not only about motors, software and kinematics.

Between the motor and the joint there is a fundamental element:

the mechanical transmission.

Its architecture directly influences torque, stiffness, inertia, backlash, vibration, accuracy, payload and dynamic behaviour.

Main transmission families used in robotics include:

• Gear trains – robust and stiff torque transmission.

• Planetary gearboxes – compact, efficient and with high torque density.

• Cycloidal / RV reducers – high stiffness, low backlash and high load capacity, often suited to the main robot axes.

• Harmonic / Strain-Wave drives – compact, lightweight and with very low backlash, often used where low inertia is important.

• Timing belt & pulley systems – transfer motion between separated shafts and allow more flexibility in motor positioning.

• Direct drive – eliminates intermediate transmission stages, but requires high-torque motors.

A useful distinction is also between robot axes.

J1–J3 mainly position the arm and normally manage higher loads and moments, so torque, stiffness and load capacity are critical.

J4–J6 mainly control wrist orientation, where compactness, low mass and low inertia become increasingly important.

Where does the AR4-MK5 fit?

The AR4-MK5 uses a hybrid mechanical transmission architecture rather than a single transmission technology.

During assembly, several axes clearly show a chain based on:

Stepper Motor → Pulley → Timing Belt → Pulley → Shaft / Joint

combined with couplers, shafts, keys/keyways, bearings and mechanical reduction stages.

So the AR4 can be associated mainly with a:

Timing Belt & Pulley + Mechanical Reduction architecture.

This allows the motor to be positioned away from the joint and the transmission ratio to be adapted to the required motion.

The basic principle is:

Motor: high speed / lower torque → Reduction → Joint: lower speed / higher torque

or, in simplified form:

Tjoint ≈ Tmotor × i × η

where i is the reduction ratio and η the transmission efficiency.

This is also why small assembly details matter.

Incorrect belt tension, coupler tightening, key fit, shaft alignment, bearing preload or backlash can later appear as:

lost motion, vibration, reduced repeatability or positioning errors.

Key takeaway

Same 6-axis architecture does not mean the same mechanical behaviour.

Robot performance depends not only on software and kinematics, but also on what happens mechanically between each motor and joint.

For PaMiLoCELL, assembling the AR4-MK5 is therefore also a practical way to understand how belts, pulleys, shafts, bearings and reduction systems work together to create robot motion.

#PaMiLoCELL #PaMiLoLAB #AR4MK5 #Robotics #IndustrialRobotics #RobotMechanics #MechanicalTransmission #TimingBelt #Gearbox #MotionControl #RobotKinematics #MechanicalEngineering #Automation

Another AR4-MK5 integration step: building a reliable moving cable harness.

In robotics, wiring is not only about electrical continuity. It also means managing flexibility, electromagnetic compatibility, mechanical stress and maintainability.

For the AR4-MK5, shielded and stranded continuous-flex CAT5/6 is used for encoder, Hall/limit-switch and other low-level signals. The outer jacket is removed where required so the twisted pairs can be separated and routed through the arm.

Key points learned during assembly:

1) EMI/RFI shielding protects sensitive feedback signals from noise generated by stepper motors and switching electronics.

2) Braided sleeve provides mechanical protection and cable organization, but does not replace electromagnetic shielding.

3) 20 AWG flexible silicone wires are used for motor phase extensions, keeping power and signal wiring clearly differentiated.

4) Controlled service loops allow the cable to move without transferring unnecessary stress directly to terminals or solder joints.

The first soldering operations are now being performed directly on the assembled robot, where real constraints such as access, routing, bend radius and moving joints become part of the engineering problem.

A further challenge is the use of Sn100 lead-free solder.

Compared with Sn63/Pb37, Sn100 requires higher working temperature and different process control. My goal is therefore not to assume that Sn100 is inherently better, but to engineer the harness so the soldered area remains outside the main cyclic bending zone. With Sn99.3/Cu0.7 at 370 °C I get more uniform wetting and a thinner bead. When soldering with 67/37, a small ball forms at 340°C, which I find difficult to remelt using flux and a soldering iron.

Initial pull tests showed no cable damage. The next reliability measures are based on:

large bend radius + service loop + heat-shrink + strain relief + controlled cable routing

The objective is to keep the rigid soldered zone mechanically protected while maintaining stable, low-resistance electrical continuity during robot motion.

This is exactly the kind of engineering challenge I want PaMiLoCELL to address: combining mechanical design, electrical integration, EMC awareness, reliability and future AI-based automation in one practical robotic platform.

#PaMiLoCELL #PaMiLoLAB #Robotics #IndustrialAutomation #EMC #EMI #RFI #CableHarness #Soldering #RobotWiring #ElectricalEngineering #ReliabilityEngineering #Engineering #LearningByDoing

During the AR4-MK5 assembly and MK4 → MK5 upgrade, one of the most useful lessons came from something apparently simple: fitting a keyed shaft.

For those unfamiliar with robot mechanics, J1 and J2 are the first two robot joints. J1 rotates the robot around its base; J2 is the shoulder joint that raises and lowers the arm.

During J2 assembly, the 14 mm gearbox shaft with parallel key would not enter its mating hub correctly.

The first instinct could be to apply more force. Instead, the problem became a practical troubleshooting exercise:

Where is the interference actually coming from?

The checks included shaft diameter, key height, keyway alignment, possible burrs and the overall shaft-plus-key dimension measured with a caliper.

The key was first pressed fully into its seat using a bench vise with protective tape. When a small residual interference remained, the key was carefully filed in very small increments, repeatedly checking the fit.

Only in this kind of exceptional situation, once the geometry and alignment are verified, a soft rubber mallet can help progressively seat the gearbox assembly, never striking the shaft or key directly.

Another important lesson came from the AR4 instructions: fully inserted does not mean eliminating every visible gap. When J2 is correctly installed, about 2 mm remains between the gearbox shoulder and the J2 tension ring.

The same reasoning applies to bolts and nuts. Fasteners should clamp correctly aligned components, not pull misaligned parts into position. Threads should be started by hand and tightened progressively, using an alternating sequence when several bolts clamp the same flange.

The MK4 → MK5 upgrade also reinforced the importance of understanding all locking points, set screws and mounting fasteners before applying extraction or assembly force.

Where specified, threadlocker must also be selected correctly: the AR4 instructions recommend low-strength threadlocker for screws 4 mm and larger, but not for 3 mm and smaller screws.

A future PaMiLoCELL upgrade will introduce torque wrenches, allowing critical fasteners to be tightened with controlled and repeatable torque rather than relying only on manual feel.

The workflow I take from this experience is:

inspect → measure → understand → correct → align → seat → tighten

Precision assembly is not about applying more force. It is about understanding dimensions, alignment, contact surfaces, interference and clamping forces.

#PaMiLoCELL #PaMiLoLAB #Robotics #MechanicalEngineering #RobotAssembly #AR4MK5 #PrecisionAssembly #Fasteners #TorqueControl

The AR4-MK5 mechanical assembly is progressing through the main transmission elements: motor, bearings, timing pulleys/gears and belts.

I approached this phase from a Quality Engineering perspective, treating each assembly step as a small Quality Gate.

Motor: Check fastening, shaft/pulley interface and cable routing. Any misalignment introduced here can later affect belt tracking and bearing loads.

Bearings: Tapered roller bearings require different installation methods:

• Outer race → aluminium housing: heat the housing uniformly to slightly expand the seat, then press the race squarely into position. Never hammer it.

• Inner bearing cone → shaft: warm the bearing/cone assembly to slightly expand the inner ring before fitting it onto the shaft. If force is required, apply it only to the inner ring, never through the rollers.

After pressing the outer race, verify seating depth at 0° / 90° / 180° / 270° with a digital caliper. This is a practical QC check of uniform seating and parallelism, while also considering measurement repeatability.

Thread locking / retaining compound

• LOCTITE 222: low-strength threadlocker for removable threaded fasteners, used on M4 and larger screws as specified by the AR4 manual.

• LOCTITE 648: retaining compound for cylindrical fits; use only if an outer race is slightly loose in its housing, not as a general installation aid and not normally between inner ring and shaft.

Gear / pulley: Check seating, axial position and pulley alignment. Small offsets can cause belt tracking issues and additional bearing loads.

Belt: Tension must prevent backlash and tooth skipping without overloading shafts, bearings or structures.

Final QC: Before closing the covers, manually check:

smooth rotation, correct belt tracking, no abnormal friction, controlled preload, no excessive play and no interference.

The key lesson is that:

Motor → Bearing → Gear/Pulley → Belt

should also be a sequence of Quality Gates.

Detecting an incorrect bearing seat, pulley misalignment or excessive belt tension immediately is far easier than troubleshooting the defect after the robot is fully assembled.

#PaMiLoCELL #PaMiLoLAB #AR4MK5 #Robotics #QualityEngineering #RobotAssembly #Bearings #TimingBelt #LessonsLearned

The AR4-MK5 wiring phase is progressing, with particular attention to the J1 stepper motor, digital drivers and control architecture inside the robot’s compact electrical cabinet.

The ground floor of the micro-cabinet is dedicated to the digital stepper drivers. Each driver effectively interfaces two different domains:

Power side: supplies the controlled phase currents to the stepper motor windings. Correct identification and connection of the motor phases, A+, A−, B+, B−, is essential for proper torque generation and rotation.

Command side: receives low-power control signals such as PULSE, DIRECTION and ENABLE from the controller. These signals define when the motor moves, in which direction and whether the driver output stage is enabled.

This separation between power wiring and control wiring is an important concept in the robot architecture: the driver acts as the interface between low-power digital commands and the considerably higher motor phase currents.

The first floor, immediately above the drivers, is dedicated to the Teensy 4.1 control board. The Teensy is connected to the digital drivers for motion commands, to selected motor feedback/sensor wiring, and will provide the interface toward the external PLC through Modbus RS-485.

The resulting control chain can therefore be summarized as:

PLC → Modbus RS-485 → Teensy 4.1 → STEP/DIR signals → Digital Drivers → Stepper Motors

During assembly, some practical wiring lessons also became clear. Small details can have a significant impact on reliability:

When stripping thin conductors, even a small nick in the copper can create a fatigue initiation point, especially where the cable is repeatedly flexed.

Moving sections should use stranded / continuous-flex conductors whenever possible, rather than rigid solid-core wires.

With screw terminals, the conductor must actually enter between the clamping plates. A wire can appear correctly positioned while remaining outside the clamp.

A simple pull test after tightening each terminal is an effective verification step.

Ferrules can improve termination quality for very small stranded wires, provided that the correct size is selected.

Some encoder conductors available on the J1 motor are not required by the standard AR4 configuration, showing the importance of understanding each signal before cutting or terminating the cable.

This phase is not only about completing the wiring. It is also about understanding how power electronics, motion control and communication layers interact inside a compact robotic system.

Next step: complete the internal wiring and progressively validate the control chain from Teensy to the motor drivers.

#PaMiLoCELL #PaMiLoLAB #AR4MK5 #Robotics #StepperMotor #MotionControl #Teensy41 #Modbus #RS485 #IndustrialAutomation #ElectricalEngineering #RobotControl