These engagements illustrate how Cluster A perception work and Cluster B motion autonomy combine in production contexts. Client names are anonymised where NDAs apply; technical decisions are described so your team can evaluate fit.
A mid-size packager needed to handle varying carton sizes from a single inbound pallet without reprogramming for every SKU rotation. We deployed a depth-assisted perception pipeline with grasp ranking trained on their actual corrugate textures — not synthetic meshes.
Six-week pilot in our bay, two weeks on-site commissioning. Outcome: stable pick cycle within agreed tolerance bands across twelve tested SKUs, with a calibration checklist maintenance adopted for weekly checks.
Force-limited gripping and joint-space paths for aluminium billets with ±0.2 mm placement requirement. Motion primitives included door-interlock coordination with the machine tool and explicit fault branches for mis-pick and air-loss events.
We integrated ROS 2 nodes with the existing Fanuc interface via a guarded middleware bridge. Operators received laminated recovery cards mapped to each fault code.
Cluster A + B
Multi-AMR milk-run — Toronto assembly campus
Three autonomous mobile robots moving kits between warehouse, kitting and two assembly halls. Cluster A work covered fiducial fallbacks and lidar drift monitoring; Cluster B covered waypoint graphs, intersection priority rules and MES-triggered dispatch.
Traffic rules respected pedestrian crossings and existing AGV contracts. Fleet reached steady-state duty after hypercare window with documented KPI dashboard shared with operations leadership.
Surface defect detection on stamped brackets under shifting overhead lighting. We built a segmentation pipeline with controlled exposure hardware and a retraining procedure tied to their quality team's label workflow.
False-reject rate reduced to contract threshold without slowing the line. Perception models and acceptance thresholds are versioned in the client's Git repository with RoboDrive runbooks for rollback.
Local context, national reach
Liberty Village gives us fast access to GTA manufacturers, film-stage fabricators and warehouse operators who need pilots without flying consultants in from abroad. We still take on projects across Ontario, Quebec and Western Canada when remote scoping and scheduled site blocks make sense. Time zones, VPN access and spare-part logistics are planned upfront — not improvised after kickoff.
Visitors to Hanna Avenue see the control bay as clients experience it: cable trays labelled, safety mats worn from real cycles, whiteboards with planner parameters still marked from yesterday's iteration. That transparency is intentional. You should know how we work before you invite us onto your floor.
Every snapshot above includes handover artefacts: scoped assumptions, sensor BOM rationale, planner parameter sheets and operator training notes. We do not deliver demos that collapse when conveyor speed changes — acceptance criteria are agreed before build starts.
Want a feasibility read on a similar challenge? Book a robotics pilot and reference the cluster that best matches your station or fleet.
Your line deserves the same rigour
Share your cycle data and constraints. We will tell you honestly whether a pilot, retainer or architecture review is the right first step.
