Time To React (TTR)¶
Description¶
The TTR metric [Hillenbrand2006] [Tamke2011] approximates the latest time until a reaction is required by aggregating the maximum TTM metric over a predefined set of maneuvers \(M\), i.e.
For example, as a set of maneuvers, one might select \(M = \{\)‘brake’, ‘steer’, ‘kickdown’\(\}\).
Properties¶
Run-time capability¶
Yes
Target values¶
Depends on maneuver set \(M\)
Subject type¶
Optimal for road vehicles (automated and human), sub-optimal for VRUs
Scenario type¶
Overlapping predicted trajectories for a significant time span in the scenario
Inputs¶
Static/dynamic objects and their state (pose, shape, etc.) at time t, set of maneuvers \(M\) and their MMs
Output scale¶
\(\{-\infty\} \cup [0,\infty]\), time (s), ratio scale
Reliability¶
High, under the assumption that collisions can be reliably predicted
Validity¶
High, claimed to be ‘an adequate metric to assess the criticality […] since it directly relates to the driver’s possible actions’ [Hillenbrand2006], but also dependent on maneuvers and collision prediction
Sensitivity¶
Comparable to TTM, as, due to selection of the maneuver with the maximal TTM, the same reasoning applies
Specificity¶
Highly increased compared to TTM due to consideration of multiple maneuvers
Prediction model¶
Time window¶
Unbound, but usefulness depends on DMM
Time mode¶
Linear time for \(|M|=1\), branching time for \(|M|>1\)