Hi Scenic maintainers,
URML (urml.dev) is a small, Apache-2.0 language for describing robot intent: an English sentence becomes a typed primitive, validated against a robot's declared capabilities and a safety envelope, then dispatched. I am reaching out because Scenic and URML are both specification languages, on complementary axes, and that feels worth exploring out loud.
URML specifies what the robot is meant to do and what it can do. Scenic specifies the world it is placed in. Put together, they describe both halves of a test: a URML behavior as the agent under test, a Scenic scenario as the environment that exercises it. To be clear, this is a peer-and-compose idea, not URML trying to sit above Scenic.
Two real questions. First, could a URML behavior serve as the ego/agent specification a Scenic scenario drives, and where would the interface naturally sit? Second, a Scenic scenario makes assumptions about the agent (its dynamics, its sensors); could those stay consistent with a URML capability manifest rather than being restated, so the two specs cannot drift?
Full write-up: https://github.com/URML-MARS/URML/blob/main/docs/rfcs/0366-scenic-outreach.md
Thanks for Scenic; a readable probabilistic scenario language raised the bar for how we talk about test environments.
Ido Yahalomi (URML, greenvh@gmail.com)
AI-assisted prose, maintainer-reviewed before posting (see VIBE.md). Human-only correspondence available on request.
Hi Scenic maintainers,
URML (urml.dev) is a small, Apache-2.0 language for describing robot intent: an English sentence becomes a typed primitive, validated against a robot's declared capabilities and a safety envelope, then dispatched. I am reaching out because Scenic and URML are both specification languages, on complementary axes, and that feels worth exploring out loud.
URML specifies what the robot is meant to do and what it can do. Scenic specifies the world it is placed in. Put together, they describe both halves of a test: a URML behavior as the agent under test, a Scenic scenario as the environment that exercises it. To be clear, this is a peer-and-compose idea, not URML trying to sit above Scenic.
Two real questions. First, could a URML behavior serve as the ego/agent specification a Scenic scenario drives, and where would the interface naturally sit? Second, a Scenic scenario makes assumptions about the agent (its dynamics, its sensors); could those stay consistent with a URML capability manifest rather than being restated, so the two specs cannot drift?
Full write-up: https://github.com/URML-MARS/URML/blob/main/docs/rfcs/0366-scenic-outreach.md
Thanks for Scenic; a readable probabilistic scenario language raised the bar for how we talk about test environments.
Ido Yahalomi (URML, greenvh@gmail.com)
AI-assisted prose, maintainer-reviewed before posting (see VIBE.md). Human-only correspondence available on request.