Fleet-Wide Pick Intelligence
One training session.
Every arm. By next shift.
Automcore teaches your robot fleet to pick any new SKU without individual arm programming — one session propagates to the entire fleet by morning.
The Problem
Per-arm programming is an O(robots × SKUs) problem
Every new product in your warehouse shouldn't require a teach-in session on every arm in your fleet. It scales terribly.
The old way
Per-arm teach-in
- Each robot arm requires individual programming for every new SKU
- 12 arms × 50 new SKUs/quarter = 600 teach-in sessions annually
- Downtime per changeover ranges from 4–12 hours per arm
- Fleet knowledge is siloed — arms can't share what they learn
- Programming costs scale directly with fleet size
The Automcore way
Train once. Propagate everywhere.
- One operator demonstration on a single arm
- Automcore compresses and distributes the grasp model fleet-wide
- Every arm validates and activates the new SKU within hours
- Fleet knowledge is shared — one session scales to all arms
- Programming cost is constant regardless of fleet size
Process
Three steps from training to full fleet
No cloud dependency. No integration complexity. Just a facility network and one edge node.
STEP 01
Capture: one arm, one session
An operator guides a single arm through the pick sequence for a new SKU. Automcore's 3D point cloud and force-torque sensor data capture object pose, gripper approach trajectory, and grasp geometry in a single demonstration. No teach pendant required.
STEP 02
Compress and distribute fleet-wide
The learned grasp model is compressed using knowledge distillation — reducing transfer size by 94% — then pushed to every arm controller via the facility LAN. No cloud round-trip. No WAN dependency. Distribution is staggered so the propagation window doesn't saturate your network switch.
STEP 03
Per-arm validation before production
Each arm runs the received model through a local pick simulation, checking grasp confidence against the required threshold. Arms that pass begin production picking. Arms that fall below threshold are flagged for a follow-up session — not silently put into production.
Platform
Monitor every arm, every SKU, every shift
The Automcore fleet dashboard gives your operations team live visibility into propagation status, pick rates, and arm health.
Compatibility
Works with the arms already in your facility
Automcore is a software layer, not a robot manufacturer. We connect to your existing arm controllers — no hardware replacement, no proprietary end effectors, no facility downtime to install.
In the Field
Deployed across warehouse operations in the Inland Empire and beyond
Representative deployments from active Automcore installations.
Regional distribution center, Inland Empire, CA
8 UR10e arms running mixed-SKU depalletizing across three product categories — consumer electronics, personal care, and dry food. Before Automcore, new seasonal SKUs required 2 hours of teach-in per arm. First 180 SKUs trained and propagated fleet-wide in 90 days.
E-commerce fulfillment operator, Phoenix, AZ
14-arm fleet of FANUC CR-14iA cobots handling high-velocity SKU turnover for peak season. A new product line of 32 SKUs arrived mid-October with a 48-hour go-live requirement. Automcore propagated the full cohort to all 14 arms inside the maintenance window.
Third-party logistics provider, Southeast US
6 FANUC CR-35iA arms (35 kg payload) handling client onboarding for a new retail account. The client's product catalog included 78 new SKUs at launch. Previously, the 3PL's robot programmer required 3 weeks to onboard a catalog that size. With Automcore: one session per SKU, fleet propagation overnight.
Your fleet can learn a new product by tomorrow morning.
One training session. Overnight propagation. Every arm ready by first shift.