INDRBT / ENGINEERING

INDRBT / Reading room

Design decisions

Digital engineering model. Payload and repeatability are design targets. All ordinary guide reading, figures and PDFs work here. Links marked Source file lead to the public engineering repository.

Adapted for browser reading from the project’s illustrated documentation. Numeric results and engineering scope are unchanged. Reproduction notes are optional; the robot and board viewers run from the navigation above.

On this page
Decision Reason Alternative / tradeoff
Python and Build123d primary CAD Same language as analysis; explicit parametric BRep export Manual CAD harder to regenerate; OpenSCAD duplicates geometry
Original Z-Y-Y-X-Y-X chain Two positioning links and three wrist rotations; all transforms inspectable Offset wrist gives orientation-dependent reach and numerical IK complexity
320/280 mm link lengths Near-balanced position workspace with reduced distal span Longer forearm increases elbow torque and inertia
Hollow aluminum links and service covers Lower structural mass with accessible cabling and actuator interfaces Simplified exterior does not specify complete manufacturing tolerances
Numerical bounded IK with SO(3) residual Handles this non-spherical offset wrist and joint limits Local convergence; multiple starts improve robustness without completeness proof
Quintic default joint motion Rest-to-rest velocity and acceleration endpoints Linear interpolation has endpoint velocity discontinuity; educational comparison only
Opposed lead-screw nuts for gripper Symmetric positive jaw travel, simple force/lead calculation Backlash, efficiency, wear and actual threads remain conceptual
Explicit lumped drive masses Exterior CAD does not model copper, magnet, bearing and gear internals CAD material volume is reported separately, not silently substituted
Analytical beam model Transparent calculations reproducible without pretending FEA Does not resolve contact, holes, bearings, fasteners or fatigue
Diagonal inertia/damping PID model Clear feedback and saturation experiment No claim of full manipulator dynamics or physically tuned gains
Low-voltage joint/sensor interface Useful meaningful electronics scope without fictional power inverter External controller and certified drive still required
Four-layer conceptual interface PCB Separate inner ground/3.3 V planes and outer signal layers simplify returns and routing Greater fabrication complexity than a two-layer board; no EMC or physical qualification claim
Full-pair representative CAD screening Selected pairs missed adjacent shoulder/wrist interfaces; broader classifications and exact intersections found the real defects Sampled configurations still do not certify every continuous joint motion
Numbered presentation folders and retained source subtree User can find models, reports and electronics together while reproduction scripts retain their source-relative paths Presentation copies must be refreshed and hash-checked after regeneration
Repository visibility Initially private because public visibility was not specified; owner explicitly requested public visibility on 21 October 2026 Current repository and inspection release are public; original history and the v1.2.0 engineering release are retained

The journal records actual iterations and failures. The validation matrix marks partial evidence explicitly.