---
title: Autonomy — condition 02 of six for commercial robotics
canonical: https://botzgroup.co/conditions/autonomy/
owns: The autonomy condition for commercial robotics
updated: 2026-08-28
---

Condition 02 of 6 Autonomy The testCan it finish a task without a human in the loop at every step? Teleoperation is a demonstration. Unattended completion, with recovery from the ordinary failures of a real environment, is a business. As at 2026-08-28 · Industry average 2.17 / 3 · Supplied by Models and simulation § 01Where the industry stands Averaged across the six categories on the Bench, autonomy scores 2.17 of three. Average2.17 / 3Across six deployment categories Strongest3 / 3Fixed industrial arms Weakest1 / 3General-purpose humanoids CategoryScoreReasoning Fixed industrial arms 3 / 3Complete within the cell. The cell exists precisely so the environment cannot surprise the machine, which is autonomy purchased with civil engineering.Warehouse mobile robots 2 / 3Navigation and routing are unattended; exception handling still routes to a human, and exception rate is the real operating metric nobody publishes.Consumer and social robots 2 / 3Conversational autonomy is strong; physical autonomy is limited and mostly not the point of the product.Quadruped inspection platforms 2 / 3Repeatable inspection routes run unattended; anything off-route needs an operator. Autonomy is a function of route familiarity, not of capability.General-purpose humanoids 1 / 3Improving quickly from a low base. Most publicly shown work is either teleoperated, heavily scripted, or performed in a environment arranged for it.Simulation-trained policies 3 / 3The strongest autonomy result in robotics. Policies trained in simulation now transfer to hardware for whole task classes that resisted hand-written control for thirty years. Studio assessment, dated 2026-08-28. Rubric and method: the six conditions. Safety framework referenced throughout: ISO 10218. § 02What supplies it Models and simulation This condition is a machine problem rather than an arrangement problem, and the industry is solving it. It is one of only two where a robotics company can hold a durable engineering advantage — which is the argument for concentrating effort here and adopting the rest. § 03The other five 012.17/3 EmbodimentCan it physically act where the work actually is? Hardware 030.67/3 AuthorityIs it permitted to act, by whom, and within what scope? flashyID 040.17/3 SettlementCan it be paid for work, and pay for what it consumes? Flashy Ledger 051.33/3 AccountabilityWhen it damages something, who answers — by name? The AAO charter 061.83/3 FleetCan one operator run a hundred of them? FlashyOS
