Scaling Safety Margins with Velocity and Momentum

This direction is split into other directions: speed-dependent headway constraint. Its last version stays readable and citable here.

Version 1 of 199 papers136 conceptsfrozen 2026-10-04

Traditional robotic safety relies on "keep-out zones"—static buffers that prevent a machine from getting too close to a human or obstacle. However, a fixed 2-meter buffer is overkill for a robot creeping at a snail's pace and dangerously insufficient for a vehicle traveling at highway speeds.

To bridge this gap, researchers are moving toward velocity-dependent safety constraints. The core intuition is that safety is not a spatial property, but a kinetic one: a state is only "safe" if the system has the time and braking power to reach a stop before a collision occurs.

The Geometry of Braking Physics

The most intuitive way to implement dynamic safety is to scale the size of a robot's "personal space" based on its current speed. In maritime navigation, Yoshiki Miyauchi and colleagues demonstrated this by using an elliptical "dynamic ship domain." As the vessel speeds up, the ellipse stretches forward, reflecting the increased distance needed to maneuver or stop.

This principle applies equally to land vehicles. Wei Xiao and Christos G. Cassandras argue that static safety buffers fail because they ignore the linear relationship between velocity and reaction distance. By enforcing a "time gap" rather than a "distance gap," the system ensures that inter-vehicle spacing naturally expands at high speeds and contracts in slow traffic, maintaining a constant safety margin in the time domain.

From Proximity to Closing Speed

Distance alone tells you how close you are; it doesn't tell you how fast you are approaching disaster. Changliu Liu and Masayoshi Tomizuka pioneered the Robot Safe Interaction System (RSIS), which uses safety metrics that account for both distance and relative velocity.

This approach produces smoother, more proactive trajectories. If a human is walking parallel to a robot, the relative velocity is low, and the robot can continue its task. If the human suddenly turns toward the robot, the closing speed spikes, triggering an immediate, proactive detour. Guanglong Du and Shuaiying Long further refined this by distinguishing between fast approaches (requiring retreat) and slow movements (allowing for efficient bypasses), optimizing the trade-off between safety and productivity.

Formalizing Safety with Control Barrier Functions

To provide mathematical guarantees of safety, researchers use Control Barrier Functions (CBFs). A CBF defines a "safe set" of states and ensures the controller never exits that set.

However, standard CBFs can be overly conservative, often "freezing" the robot when it gets near an obstacle. Manan Tayal and Shishir Kolathaya addressed this by combining CBFs with Collision Cones. By scaling the safety intervention based on the magnitude of relative velocity, their "C3BF" framework only restricts the robot when a collision is imminent at high speeds, allowing for much more fluid movement in tight spaces.

Similarly, Bolun Dai and Farshad Khorrami developed time-varying CBFs that incorporate the time-derivative of the safe set. This allows the robot to "see" the velocity of moving obstacles and inflate its safety boundaries only in the direction of the obstacle's motion, providing a computationally efficient way to account for physical actuation limits.

Momentum as the Ultimate Safety Metric

In heterogeneous environments where e-scooters share the road with heavy SUVs, velocity-based scaling isn't enough—you must also account for mass. Robert N. Shorten and Wynita M. Griggs propose using momentum ($p = mv$) as the primary safety metric.

By setting vehicle-specific speed limits based on mass, the system ensures that the kinetic energy involved in any potential collision remains within survivable limits. Under this framework, a heavy truck might be restricted to 20 km/h in a pedestrian zone, while a lightweight bicycle is permitted to travel faster, as its lower mass results in a lower "impact severity" at the same speed.

Why This Works: Reducing Conservativeness

The shift from static to dynamic safety margins works because it aligns the robot's "caution" with the physical reality of the environment.

  • High Speed/High Risk: The system expands its safety buffers, initiating evasive maneuvers early to account for long braking distances.
  • Low Speed/Low Risk: The system shrinks its buffers, allowing for high-precision maneuvers and close-quarters interaction that would be impossible under rigid, distance-only constraints.

As Han Wang and colleagues note in their work on Probabilistic Occupancy Risk Assessment, scaling boundaries based on "stopping sight distance" ensures that autonomous systems mimic human-like safety behavior—being most cautious exactly when their kinetic state makes them most dangerous.

Papers in this version

Showing 60 of 99 papers.

History & citationVersion 1 · 2026-10-04

Split into other directions · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_scaling_safety_margins_with_velocity_and_84310ae7,
  title = {Scaling Safety Margins with Velocity and Momentum},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/scaling-safety-margins-with-velocity-and-momentum/rdr_84310ae747e94a3a84028e10c3628515},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_scaling_safety_margins_with_velocity_and_84310ae7_v1,
  title = {Scaling Safety Margins with Velocity and Momentum},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/scaling-safety-margins-with-velocity-and-momentum/rdr_84310ae747e94a3a84028e10c3628515/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}