TNLROBOTICSDefine a pilot

Vision systems for autonomous local manufacturing

Inspection that holds up at production speed.

TNL Robotics engineers the camera, lighting, and product fixtures that turn an AI robot into a repeatable inspection cell.

See the inspection stack OpenArm reference

Human skill, multiplied by AI robotics.

REFERENCE PLATFORMOPENARM · 7DOF
OpenArm dual-arm robot and standardized OpenArm CellCAMERA PLANECONTROLLED LIGHTPRODUCT DATUM
OpenArm provides the motion platform. TNL focuses on the inspection layer that production depends on.

OPEN ROBOT PLATFORM

CONTROLLED VISUAL EVIDENCE

REPEATABLE QUALITY DECISIONS

01 / INSPECTION SYSTEM

Three variables.
One dependable result.

A model cannot recover information the physical system failed to reveal. We control the image before asking AI to interpret it.

01

Machine vision

Camera, lens, exposure, and trigger logic engineered around the defect—not around a catalog part.

Stable acquisition · calibrated optics · traceable images
02

Purpose-built lighting

Light geometry, spectrum, and timing designed to make the relevant surface information measurable.

Exposure-synced · application-specific · serviceable
03

Product fixturing

Jigs that constrain part pose, protect critical surfaces, and support repeatable, fast changeovers.

Datum-controlled · poka-yoke · production-ready

02 / ENGINEERING METHOD

Build the evidence path.

Every inspection cell is treated as a chain. A weak link in pose, light, optics, timing, or decision logic becomes a false reject—or a missed defect.

  1. 01

    Observe

    Capture the human process and the inspection decision.

  2. 02

    Control

    Lock part pose, camera geometry, and illumination.

  3. 03

    Inspect

    Turn visible evidence into a repeatable quality result.

  4. 04

    Act

    Let the robot sort, rework, or route the product.

CAM
LIGHT
PART
JIG / DATUM

THE PHYSICAL AI PRINCIPLE

“Control what the camera sees before optimizing what the model predicts.”
Input
Consistent, defect-relevant images
Decision
Traceable pass / fail evidence
Action
Robot routing or rework

03 / PROTOTYPE BASELINE

Concrete enough to test.

This reference configuration turns current engineering work into a measurable starting point. Final components remain application-dependent.

EXAMPLE CONFIGURATION · NOT A FINAL PRODUCT SPECIFICATION

01 · IMAGING

Basler ace 2

  • Referencea2A3536-31umPRO
  • InterfaceUSB 3
  • TriggerExposure Active

02 · ILLUMINATION

Vision ring light

  • Geometry80 / 45 mm OD / ID
  • White5700–6500 K
  • Contrast option625 nm red

03 · SYNCHRONIZATION

Exposure-timed light

  • ControlCamera GPIO
  • IsolationOptocoupled driver
  • System rail24 V DC

04 / DEPLOYMENT VALUE

Designed for the factory around the robot.

R

Repeatability

Fixture datums and controlled illumination reduce variation before software.

T

Traceability

Every decision can retain its image, timestamp, recipe, and model version.

C

Changeover

Modular jigs and inspection recipes make product variants manageable.

S

Serviceability

Replaceable lights, accessible optics, and documented setup preserve uptime.

START WITH ONE PART

Define the defect.
Then build the cell.

A useful pilot begins with part samples, a defect taxonomy, and the real line cycle time. Copy the brief and complete it with your engineering team.

No form submission. Your information stays with you.