Open problem
How does a machine know what a human really means?
nodes // human-intent
Real World Atlas / open problems
The future is not a list of answers. It is a map of unsolved problems.
Open problems in physical AI, embodied intelligence and autonomous systems. Nothing on this page is presented as solved, and no speculative answer is offered as fact.
INTENT
PERCEPTION
Open problem
How much of a scene must be understood before acting is safe?
nodes // perception
PLANNING
Open problem
How does a plan stay valid in an environment that keeps moving?
nodes // planning → world-understanding
DEXTERITY
Open problem
Why is contact-rich manipulation still harder than language?
nodes // physical-execution
NAVIGATION
Open problem
How do autonomous systems handle the long tail of physical space?
nodes // physical-execution → perception
IDENTITY
Open problem
Who is an agent, legally and technically, when it acts for someone?
nodes // identity
PAYMENTS
Open problem
Who gives an autonomous agent permission to spend money?
nodes // payments → permissions
Open problem
How does a machine purchase a real-world service end to end?
nodes // payments → orchestration
PERMISSIONS
Open problem
What are the limits of delegated authority for a physical action?
nodes // permissions
HUMAN COORDINATION
Open problem
When should execution move from machine to human?
nodes // human-loop → failure-recovery
Open problem
How do machines coordinate with humans who do not behave deterministically?
nodes // human-intent → orchestration
EXECUTION
Open problem
How should robots interact with businesses that have no machine-readable interface?
nodes // orchestration
Open problem
How does an autonomous system operate where APIs do not exist?
nodes // orchestration → physical-execution
Open problem
How does an agent choose between human and machine execution?
nodes // orchestration → human-loop
FAILURE RECOVERY
Open problem
How does physical AI recover when reality diverges from the plan?
nodes // failure-recovery
Open problem
What does a retry mean when the first attempt changed the world?
nodes // failure-recovery → world-understanding
VERIFICATION
Open problem
How does a system verify that a physical outcome occurred?
nodes // verification
ECONOMICS
Open problem
Which cost curve makes physical execution economically inevitable?
nodes // orchestration → learning
TRUST
Open problem
How do physical agents establish trust with people and businesses?
nodes // identity → verification
// Thesis 063
Intelligence withoutexecution is unfinished.
// Milan, Italy
// Real World Atlas
// Observation continues