Resources · AMR Picking Guide
The two dominant AMR picking models — compared side by side on throughput, cost, deployment time, and operational fit.
When evaluating AMRs for warehouse picking, the most consequential decision isn't which vendor to choose — it's which picking architecture fits your operation. Goods-to-Person (G2P) and Person-to-Goods (P2G) systems produce dramatically different results in throughput, required investment, facility changes, and deployment timeline.
This guide explains how each model works, where each excels, and how to identify which one your operation should be evaluating. SCxChange Hub is independent and vendor-neutral. Our goal is to help you ask the right questions before you commit.
| Factor | Person-to-Goods (P2G) | Goods-to-Person (G2P) |
|---|---|---|
| How it works | A robot escorts a human picker through the warehouse, carrying the tote and directing them to each pick location via a tablet or screen. | Robots autonomously retrieve mobile shelving pods or bins and deliver them to stationary human pick stations. Pickers never leave the station. |
| Walking eliminated | Partial — robot carries the tote, but pickers still walk between locations. | Nearly complete — walking is reduced to near zero. Pickers stay at a fixed station. |
| Picks per operator/hour | 100–200 (vs. 60–100 manual) | 300–600 (vs. 60–100 manual) |
| Productivity uplift | 50–100% vs. manual picking | 200–400% vs. manual picking |
| Facility changes required | Minimal — works in existing rack layouts. No reslotting required. | Significant — requires mobile shelving pods or bin systems; existing static racking must be replaced or supplemented. |
| Capex requirement | Lower — fewer infrastructure changes. Deployment typically $300K–$1M for a starter fleet. | Higher — pod systems and pick stations add cost. Typical deployments start at $750K–$2M+. |
| Deployment timeline | 4–10 weeks typical. Some vendors achieve go-live in under 30 days. | 3–6 months typical. Pod system setup, mapping, and integration take longer. |
| Best SKU profile | Medium-to-high variety, moderate velocity. Works well with traditional pick paths. | High velocity, high order volume, repetitive picks. Most efficient with a large, consistent SKU set. |
| Space efficiency | Same as existing layout — no density improvement. | Higher density — pods can be stored tightly because robots navigate the grid, not humans. |
| Scalability | Add robots incrementally. Easy to scale up or down. | Scalable but more capital-intensive per increment. More planning required to expand pod inventory. |
| Best fit operation | Brownfield retrofits, 3PLs, operations with mixed workflows, seasonal peaks. | Greenfield DCs, high-volume e-commerce, operations with consistent high-velocity SKUs. |
| Vendor examples | Locus Robotics, 6 River Systems (Zebra), GreyOrange Ranger, Fetch Robotics | Geek+, HAI Robotics, Amazon Robotics, Brightpick, AutoStore (fixed grid variant) |
Within P2G and G2P there are distinct orchestration models — each with different performance profiles.
Follow Me (P2G)
The robot and picker move together as a unit. The picker is assigned to one robot for their shift. The robot carries the tote, shows pick instructions on a screen, and navigates between locations. Simple, intuitive, and fast to deploy.
Best for: Teams new to robotics, brownfield retrofits, operations with low tolerance for process change
Find Me / Swarm (P2G)
Multiple robots roam a zoned area independently. Pickers move through the zone and pick up whichever robot has work available, usually indicated by a flashing light or audio cue. Higher throughput than Follow Me because robots aren't idle while pickers scan items.
Best for: Higher-volume operations, larger facilities, operations with multiple pick zones running simultaneously
Meet Me (P2G)
An evolution of the swarm model. The WES or orchestration software directs both pickers and robots independently, coordinating them to arrive at a pick location at the same time. Pickers have their own mobile app and move through the warehouse on an optimized path — the robot meets them at the pick location.
Best for: High-performance operations that want to maximize both robot and human utilization simultaneously
Goods-to-Person (G2P)
Robots retrieve mobile shelving pods or bin totes and bring them to a fixed pick station. The human picker stays stationary and picks from each pod as it arrives. Walking is eliminated entirely. The highest-throughput AMR architecture available today.
Best for: High-volume e-commerce, large DC operations, any operation where walking time is the primary inefficiency
Fully Autonomous Picking
Mobile manipulators — robots with an arm — perform the entire pick without human involvement. Navigation, item identification, grasp, and placement are all autonomous. Still maturing as of 2026, with best results on well-defined SKU sets.
Best for: Operations with high-volume, consistent SKUs willing to pilot emerging technology. Early adopters only.
Real-world performance data from deployed AMR systems (2024–2026 fleet data).
| Picking Model | Picks/Operator/Hour | Labor Reduction | Order Accuracy |
|---|---|---|---|
| Manual picking (no robots) | 60–100 | Baseline | ~99.0% |
| P2G AMR (Follow Me) | 100–180 | ↓ 30–40% | ~99.5% |
| P2G AMR (Swarm / Find Me) | 150–250 | ↓ 40–55% | ~99.7% |
| Goods-to-Person AMR | 300–600 | ↓ 50–70% | ~99.9% |
| Fully Autonomous Picking | 200–400 | ↓ 70–90% | ~99.5%+ |
Source: SCxChange Hub analysis of published vendor performance data and industry deployment reports, 2024–2026.
1. What is your current pick rate and where is the bottleneck?
If walking accounts for 50%+ of shift time, G2P will deliver dramatically more value. If your pickers are slow on scan/pick, P2G may be sufficient.
2. Can you modify your facility and storage system?
G2P requires replacing static racking with mobile pods. If your facility isn't suited for that, P2G is the practical path.
3. What is your available capex and payback target?
P2G has lower entry cost and faster payback. G2P delivers more ROI per robot but requires more upfront investment.
4. How variable is your order profile?
Highly variable orders (different SKUs, lots of one-offs) favor P2G. High-velocity, repetitive SKU sets get the most from G2P.
5. What's your deployment timeline?
Need live in 60 days? P2G is realistic. G2P at 3–6 months minimum. Factor this into your peak season planning.
Many AMR vendors — including Locus Robotics, 6 River Systems, and others — offer Robot-as-a-Service pricing: a monthly per-robot subscription that eliminates the large upfront hardware purchase. RaaS typically includes hardware, software, maintenance, and support in one monthly fee. It's particularly well-suited for 3PLs with variable volume, smaller DCs, or operations that want to pilot before committing to a full fleet purchase. Expect RaaS pricing in the range of $2,500–$5,000 per robot per month depending on vendor and contract terms.
SCxChange Hub provides independent guidance on AMR selection — no vendor commissions, no preferred platforms. We help you figure out which architecture and which vendors make sense for your specific volume, SKU profile, and facility.
Independent, vendor-neutral advice. No pressure — we respond within 2 business days.
Also explore these related resources from SCxChange Hub