Atlas // knowledge graph

Nothing here is a list of articles.

68 entities, 163 relations. Research dossiers, open problems, categories, execution stages and the research systems working on them — traversed one node at a time. Any node you focus becomes its own address.

RESEARCH DOSSIERPERMISSION TO ACTHUMAN INTENTAUTONOMOUS MOBILITYMACHINE COMMERCEIDENTITYPERMISSIONSPAYMENTSORCHESTRATIONVERIFICATIONPHYSICAL AI INFRASTRUCTUREIDENTITY · 06PERMISSIONS · 09TRUST · 18PAYMENTS · 07

RESEARCH DOSSIER // Dossier 07 // UNVERIFIED

PERMISSION TO ACT

Who gives AI permission to act in the real world?

Relations 13

Open full dossier →

RESEARCH DOSSIER · 3

EXECUTION STAGE · 5

CATEGORY · 1

OPEN PROBLEM · 4

Full relation index

  • EXECUTION STAGE

    PERCEPTION

    Reading a physical scene accurately enough to act inside it.

    Connected to 11 entities · EMBODIED AI, VISION-LANGUAGE-ACTION MODELS, SIMULATION, SENSING & TELEMETRY

  • EXECUTION STAGE

    HUMAN INTENT

    Recovering what a person actually wants from ambiguous, underspecified natural language.

    Connected to 10 entities · HUMAN-ROBOT INTERACTION, HUMAN-IN-THE-LOOP, VISION-LANGUAGE-ACTION MODELS, REAL-WORLD SERVICES

  • EXECUTION STAGE

    WORLD UNDERSTANDING

    Holding a representation of an environment that keeps changing.

    Connected to 09 entities · WORLD MODELS, SIMULATION, EMBODIED AI, EMBODIED INTELLIGENCE

  • EXECUTION STAGE

    PLANNING

    Turning a goal into an ordered sequence of physically possible steps.

    Connected to 10 entities · WORLD MODELS, ROBOT LEARNING, EMBODIED AI, ROBOT MANIPULATION

  • EXECUTION STAGE

    IDENTITY

    Who an agent is when it acts on someone else's behalf in the world.

    Connected to 08 entities · AGENTIC COMMERCE, PHYSICAL AI INFRASTRUCTURE, PHYSICAL AI INFRASTRUCTURE, IDENTITY · 06

  • EXECUTION STAGE

    PERMISSIONS

    What an autonomous system is allowed to do, and who authorised it.

    Connected to 09 entities · AGENTIC COMMERCE, HUMAN-IN-THE-LOOP, PHYSICAL AI INFRASTRUCTURE, HUMAN ↔ MACHINE

  • EXECUTION STAGE

    PAYMENTS

    How a machine pays for a real-world service under a mandate and a limit.

    Connected to 09 entities · AGENTIC COMMERCE, REAL-WORLD SERVICES, REAL-WORLD ORCHESTRATION, PHYSICAL AI INFRASTRUCTURE

  • EXECUTION STAGE

    ORCHESTRATION

    Coordinating humans, providers, APIs and machines toward one outcome.

    Connected to 17 entities · REAL-WORLD SERVICES, HOSPITALITY ROBOTICS, DELIVERY ROBOTICS, PHYSICAL AI INFRASTRUCTURE

  • EXECUTION STAGE

    PHYSICAL EXECUTION

    The moment intelligence touches matter: motion, contact, transport.

    Connected to 14 entities · ROBOT MANIPULATION, DEXTERITY, HUMANOIDS, AUTONOMOUS MOBILITY

  • EXECUTION STAGE

    FAILURE RECOVERY

    What happens when reality diverges from the plan mid-execution.

    Connected to 09 entities · ROBOT LEARNING, HUMAN-IN-THE-LOOP, DELIVERY ROBOTICS, LONG-HORIZON EXECUTION

  • EXECUTION STAGE

    VERIFICATION

    Proving the intended physical outcome actually occurred.

    Connected to 15 entities · PHYSICAL AI INFRASTRUCTURE, REAL-WORLD SERVICES, HUMAN-ROBOT INTERACTION, AUTONOMOUS MOBILITY

  • EXECUTION STAGE

    LEARNING

    Feeding real-world outcomes back into the system that caused them.

    Connected to 06 entities · ROBOT LEARNING, SIMULATION, WORLD MODELS, ROBOTICS ECONOMICS

  • CATEGORY

    HUMAN INTENT

    How machines understand what humans actually want.

    Connected to 03 entities · HUMAN INTENT, PLANNING, HUMAN INTENT

  • CATEGORY

    EMBODIED INTELLIGENCE

    How intelligence perceives and reasons inside physical environments.

    Connected to 02 entities · PERCEPTION, WORLD UNDERSTANDING

  • CATEGORY

    MANIPULATION & DEXTERITY

    How machines physically interact with objects.

    Connected to 02 entities · PHYSICAL EXECUTION, HUMANOID DEXTERITY

  • CATEGORY

    WORLD MODELS

    How intelligent systems represent and predict physical environments.

    Connected to 02 entities · WORLD UNDERSTANDING, PLANNING

  • CATEGORY

    LONG-HORIZON EXECUTION

    How machines complete tasks spanning many steps, changing state and time.

    Connected to 04 entities · PLANNING, ORCHESTRATION, FAILURE RECOVERY, LONG-HORIZON EXECUTION

  • CATEGORY

    AUTONOMOUS MOBILITY

    How machines move humans, objects and themselves through physical space.

    Connected to 03 entities · PHYSICAL EXECUTION, PERCEPTION, AUTONOMOUS MOBILITY

  • CATEGORY

    HUMAN ↔ MACHINE

    How humans and intelligent machines coordinate.

    Connected to 03 entities · HUMAN INTENT, PERMISSIONS, VERIFICATION

  • CATEGORY

    REAL-WORLD ORCHESTRATION

    How intelligence coordinates external systems, providers, APIs, humans and machines.

    Connected to 02 entities · ORCHESTRATION, PAYMENTS

  • CATEGORY

    FAILURE & RECOVERY

    What happens when physical execution does not go according to plan.

    Connected to 02 entities · FAILURE RECOVERY, FAILURE & RECOVERY

  • CATEGORY

    VERIFICATION

    How a system knows the intended physical outcome actually happened.

    Connected to 01 entities · VERIFICATION

  • CATEGORY

    ROBOTICS ECONOMICS

    Cost curves, deployment economics, labour substitution and complementarity, utilisation, scaling.

    Connected to 02 entities · ORCHESTRATION, LEARNING

  • CATEGORY

    PHYSICAL AI INFRASTRUCTURE

    The infrastructure required underneath intelligence operating in the physical world.

    Connected to 07 entities · IDENTITY, PERMISSIONS, PAYMENTS, ORCHESTRATION

  • RESEARCH SYSTEM

    ROBOT LEARNING

    Policies learned from data, demonstration and interaction.

    Connected to 03 entities · PLANNING, FAILURE RECOVERY, LEARNING

  • RESEARCH SYSTEM

    HUMANOIDS

    General-purpose embodiment in human environments.

    Connected to 02 entities · PHYSICAL EXECUTION, PERCEPTION

  • RESEARCH SYSTEM

    AUTONOMOUS MOBILITY

    Moving people, objects and machines through physical space.

    Connected to 03 entities · PHYSICAL EXECUTION, PERCEPTION, VERIFICATION

  • RESEARCH SYSTEM

    ROBOT MANIPULATION

    Grasping, placing and acting on objects.

    Connected to 02 entities · PHYSICAL EXECUTION, PLANNING

  • RESEARCH SYSTEM

    DEXTERITY

    Fine contact-rich control; hands as the hardest interface.

    Connected to 01 entities · PHYSICAL EXECUTION

  • RESEARCH SYSTEM

    HOME ROBOTICS

    Unstructured private environments with no schema.

    Connected to 02 entities · PHYSICAL EXECUTION, HUMAN INTENT

  • RESEARCH SYSTEM

    HOSPITALITY ROBOTICS

    Service execution where guests, staff and machines share a floor.

    Connected to 02 entities · ORCHESTRATION, VERIFICATION

  • RESEARCH SYSTEM

    DELIVERY ROBOTICS

    Last-metre handover between machines and people.

    Connected to 02 entities · ORCHESTRATION, FAILURE RECOVERY

  • RESEARCH SYSTEM

    EMBODIED AI

    Intelligence that reasons from inside a body in an environment.

    Connected to 03 entities · PERCEPTION, WORLD UNDERSTANDING, PLANNING

  • RESEARCH SYSTEM

    WORLD MODELS

    Learned predictive representations of physical dynamics.

    Connected to 03 entities · WORLD UNDERSTANDING, PLANNING, LEARNING

  • RESEARCH SYSTEM

    SIMULATION

    Synthetic environments, and the gap back to reality.

    Connected to 03 entities · PERCEPTION, WORLD UNDERSTANDING, LEARNING

  • RESEARCH SYSTEM

    VISION-LANGUAGE-ACTION MODELS

    Language and vision mapped directly onto action.

    Connected to 02 entities · PERCEPTION, HUMAN INTENT

  • RESEARCH SYSTEM

    HUMAN-ROBOT INTERACTION

    Legibility, trust and handover between people and machines.

    Connected to 02 entities · HUMAN INTENT, VERIFICATION

  • RESEARCH SYSTEM

    HUMAN-IN-THE-LOOP

    When execution should escalate from machine to person.

    Connected to 03 entities · HUMAN INTENT, PERMISSIONS, FAILURE RECOVERY

  • RESEARCH SYSTEM

    AGENTIC COMMERCE

    Identity, mandate and payment for autonomous buyers.

    Connected to 03 entities · IDENTITY, PERMISSIONS, PAYMENTS

  • RESEARCH SYSTEM

    REAL-WORLD SERVICES

    Providers, availability and businesses without machine interfaces.

    Connected to 04 entities · HUMAN INTENT, PAYMENTS, ORCHESTRATION, VERIFICATION

  • RESEARCH SYSTEM

    PHYSICAL AI INFRASTRUCTURE

    The layer underneath intelligence that operates in the world.

    Connected to 04 entities · IDENTITY, PERMISSIONS, ORCHESTRATION, VERIFICATION

  • RESEARCH SYSTEM

    SENSING & TELEMETRY

    The instruments that report physical state back.

    Connected to 02 entities · PERCEPTION, VERIFICATION

  • OPEN PROBLEM

    INTENT · 01

    How does a machine know what a human really means?

    Connected to 02 entities · HUMAN INTENT, HUMAN INTENT

  • OPEN PROBLEM

    PERCEPTION · 02

    How much of a scene must be understood before acting is safe?

    Connected to 01 entities · PERCEPTION

  • OPEN PROBLEM

    PLANNING · 03

    How does a plan stay valid in an environment that keeps moving?

    Connected to 03 entities · PLANNING, WORLD UNDERSTANDING, LONG-HORIZON EXECUTION

  • OPEN PROBLEM

    DEXTERITY · 04

    Why is contact-rich manipulation still harder than language?

    Connected to 02 entities · PHYSICAL EXECUTION, HUMANOID DEXTERITY

  • OPEN PROBLEM

    NAVIGATION · 05

    How do autonomous systems handle the long tail of physical space?

    Connected to 03 entities · PHYSICAL EXECUTION, PERCEPTION, AUTONOMOUS MOBILITY

  • OPEN PROBLEM

    IDENTITY · 06

    Who is an agent, legally and technically, when it acts for someone?

    Connected to 02 entities · IDENTITY, PERMISSION TO ACT

  • OPEN PROBLEM

    PAYMENTS · 07

    Who gives an autonomous agent permission to spend money?

    Connected to 04 entities · PAYMENTS, PERMISSIONS, MACHINE COMMERCE, PERMISSION TO ACT

  • OPEN PROBLEM

    PAYMENTS · 08

    How does a machine purchase a real-world service end to end?

    Connected to 03 entities · PAYMENTS, ORCHESTRATION, MACHINE COMMERCE

  • OPEN PROBLEM

    PERMISSIONS · 09

    What are the limits of delegated authority for a physical action?

    Connected to 02 entities · PERMISSIONS, PERMISSION TO ACT

  • OPEN PROBLEM

    HUMAN COORDINATION · 10

    When should execution move from machine to human?

    Connected to 02 entities · FAILURE RECOVERY, FAILURE & RECOVERY

  • OPEN PROBLEM

    HUMAN COORDINATION · 11

    How do machines coordinate with humans who do not behave deterministically?

    Connected to 03 entities · HUMAN INTENT, ORCHESTRATION, HUMAN INTENT

  • OPEN PROBLEM

    EXECUTION · 12

    How should robots interact with businesses that have no machine-readable interface?

    Connected to 02 entities · ORCHESTRATION, MACHINE COMMERCE

  • OPEN PROBLEM

    EXECUTION · 13

    How does an autonomous system operate where APIs do not exist?

    Connected to 03 entities · ORCHESTRATION, PHYSICAL EXECUTION, MACHINE COMMERCE

  • OPEN PROBLEM

    FAILURE RECOVERY · 14

    How does physical AI recover when reality diverges from the plan?

    Connected to 02 entities · FAILURE RECOVERY, FAILURE & RECOVERY

  • OPEN PROBLEM

    FAILURE RECOVERY · 15

    What does a retry mean when the first attempt changed the world?

    Connected to 04 entities · FAILURE RECOVERY, WORLD UNDERSTANDING, LONG-HORIZON EXECUTION, FAILURE & RECOVERY

  • OPEN PROBLEM

    VERIFICATION · 16

    How does a system verify that a physical outcome occurred?

    Connected to 02 entities · VERIFICATION, AUTONOMOUS MOBILITY

  • OPEN PROBLEM

    ECONOMICS · 17

    Which cost curve makes physical execution economically inevitable?

    Connected to 02 entities · ORCHESTRATION, LEARNING

  • OPEN PROBLEM

    TRUST · 18

    How do physical agents establish trust with people and businesses?

    Connected to 03 entities · IDENTITY, VERIFICATION, PERMISSION TO ACT

  • OPEN PROBLEM

    EXECUTION · 19

    How does an agent choose between human and machine execution?

    Connected to 02 entities · ORCHESTRATION, AUTONOMOUS MOBILITY

  • RESEARCH DOSSIER

    HUMAN INTENT

    When a human says what they want, does the machine actually understand the goal?

    Connected to 08 entities · HUMAN INTENT, PLANNING, PHYSICAL EXECUTION, HUMAN INTENT

  • RESEARCH DOSSIER

    LONG-HORIZON EXECUTION

    What happens after step one?

    Connected to 09 entities · HUMAN INTENT, PLANNING, WORLD UNDERSTANDING, ORCHESTRATION

  • RESEARCH DOSSIER

    FAILURE & RECOVERY

    What happens when reality says no?

    Connected to 09 entities · LONG-HORIZON EXECUTION, FAILURE RECOVERY, WORLD UNDERSTANDING, VERIFICATION

  • RESEARCH DOSSIER

    HUMANOID DEXTERITY

    Why is the human hand still so difficult to replace?

    Connected to 07 entities · FAILURE & RECOVERY, PHYSICAL EXECUTION, PERCEPTION, LEARNING

  • RESEARCH DOSSIER

    AUTONOMOUS MOBILITY

    When the driver disappears, what infrastructure remains?

    Connected to 11 entities · HUMANOID DEXTERITY, PHYSICAL EXECUTION, IDENTITY, PAYMENTS

  • RESEARCH DOSSIER

    MACHINE COMMERCE

    How does a machine buy something in the real world?

    Connected to 12 entities · AUTONOMOUS MOBILITY, IDENTITY, PERMISSIONS, PAYMENTS

  • RESEARCH DOSSIER

    PERMISSION TO ACT

    Who gives AI permission to act in the real world?

    Connected to 13 entities · HUMAN INTENT, AUTONOMOUS MOBILITY, MACHINE COMMERCE, IDENTITY

// Milan, Italy

// Real World Atlas

// Observation continues