Scribd
Upload
10 views

L126 - Autonomous Vehicle - Trajectories

Uploaded by

comodidev-store
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
10 views

L126 - Autonomous Vehicle - Trajectories

Uploaded by

comodidev-store
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

AV Trajectories:

Newtonian Mechanics
vs. The Real World
Prof. Philip Koopman

@PhilKoopman
Overview
 Limits on trajectory control
https://bit.ly/2QEOZoP
 Vehicle capability
 Environmental conditions

 Uncertainty
 About vehicle conditions
 About environment

 Managing ODD variations


 Micro-ODDs as an approach

© 2022 Philip Koopman 2


Example: Safe Following Distance
RESPONSE STOP

 Follower stops with space left behind leader (RSS example)


 Different initial speeds
 Follower initially accelerating during response time
 Different braking capabilities
 Considered safe if any gap between vehicles at rest

© 2022 Philip Koopman 3


F=MA
Not Just
A Good Idea
…

Sir Isaac Newton


It’s the Law!
4
But, Where Does the “A” Come From?
 F = MA  A = M / F
 BUT … F is limited by tire friction force

 Example: braking depends upon:


 Ability of vehicle to exert force on roadway (Ffriction)
 Driver applying full Ffriction via brakes (braking capacity)

© 2022 Philip Koopman 5


Road Conditions Affecting Braking
 Slopes
 Decreases friction AND pulls car

 Curves: ke
A bra
 Friction maintains centripetal force
 Banking (superelevation)
– Reverse bank reduces normal force
µ

 Road surface condition µ


Fnormal = mg cos θ
sinθ
g
 Dry concrete μ = 0.75 =
eθ
S lo p
l
A hil
 Snow μ = 0.2 – 0.25
 Ice μ = 0.1 – 0.15
© 2022 Philip Koopman 6
Other Factors Affecting Brake Force
 Braking capability:
 Tire capability (“sticky” tires might have μ > 1)
 Brake maximum friction (pad wear)

 Equipment condition
 Tire condition: temperature, pressure, tread
 Brake condition: hot, wet, damaged, …
 Vehicle suspension, weight distribution, …

 Braking controls https://bit.ly/2lyjq3u

 Driver leg strength and willingness to brake hard


 Braking assist force (multiplies driver leg strength)

 Aerodynamics, suspension, debris, …


© 2022 Philip Koopman 7
Epistemic Uncertainty – Vehicles
 Own vehicle weak braking (less than expected)
 Brake wear & failures
 Loss of brake assist
 High tire pressure / bald tires
 Brakes hot from recent use
 Brakes wet from recent puddle

 Other vehicle strong braking


 Braking capability for vehicle type
https://bit.ly/2jXhLE8

 Aftermarket brake upgrade?


 Aftermarket tire upgrade? Low tire pressure?
 Leg strength of lead driver to press brakes?
© 2022 Philip Koopman 8
Epistemic Uncertainty – Environment
 Road surface of own vehicle
 Might not be same as lead vehicle surface

 Road surface of lead vehicle


 Might have dramatically different friction properties
https://bit.ly/2lSBPYT

https://bit.ly/2jZqnu5

© 2022 Philip Koopman 9


Segmenting Into Micro-ODDs
 A single huge ODD leads
to poor permissiveness
 Want better performance
on a warm dry day
 Approach: break up ODDs
into pieces
 Default cautious behavior
 Prove safe trajectory for an
ODD segment
 Optimize segments based
on customer value
© 2022 Philip Koopman 10
Micro-ODD Benefits
 Turns ODD growth on its head:
 Over time: Improve permissiveness for fixed ODD size
 Operate across a diverse ODD safely (and cautiously!)
 Incrementally improve performance in high value ODD segments
 Use finer grain ODD segments for high value operational situations
– Note: important to address transition between segments
 References:
 Micro-ODD paper: https://arxiv.org/abs/1911.01207
 ODD parameter paper: https://bit.ly/33K26uA
 UL 4600
– Sections 8.2 (ODD) & 8.8 (Trajectory & Control)
© 2022 Philip Koopman 11
Conclusions
 Proofs are great, but rely upon assumptions
 In particular, about environment & behaviors
 Permissiveness vs. safety tradeoffs

?
 Proofs push uncertainty into the assumptions ? ?
?
 Uncertainty about own system
 Uncertainty about other actor behaviors ?
 Uncertainty about the environment ? ?

 You might forget the edge cases…


… but they won’t forget you!
© 2022 Philip Koopman 12

You might also like