12 Jesper - MultiDomain - Optimization

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Multidomain

Optimization of
Extreme Mobility
Vehicle
Unrestricted © Siemens AG 2018
Distribution A, Approved for Public Release
MOBILITY | SURVIVABILITY |
ROBOTICS

Distribution A, Approved for Public Release


Performance Objectives
• Develop a vehicle demonstrator that
– Minimizes the “No-Go” area
– Should be able to traverse 95% of all land terrain
– Goes faster over rough terrain than anything else
• High Level Performance Objectives
– Minimize Driver Absorbed Power
• <6 watts driver absorbed power
– Maintain Acceptable Peak Occupant Acceleration
• 2.5g max on driver Half Rounds, 4”  12”
– High Lateral/Roll Stability
– ISO double-lane change @ 65 mph
– Terrain/Mobility Performance
• 45% side slopes
• Grades
• Vertical steps
• Turning circle

– This research was developed with funding from the


Defense Advanced Research Projects Agency (DARPA).

Distribution A, Approved for Public Release 3


Suspension background

• It is well known that more wheel travel produces better ride


• Active suspension also provides better mobility
– But the drawback is the invariants in the transfer functions
– A traditional suspension can only fix one of:
• Better body control gives worse wheel control
• Better wheel control gives worse ride
• That is why you often see “Sport mode” and “Comfort mode” settings

So we are putting one suspension at the end of another

Distribution A, Approved for Public Release 4


METS Suspension Design

• METS Dual Mode Suspension Design


– High Travel (HT) active suspension for body control and to traverse large
obstacles, 72” total @ wheel center
– Low Travel (LT) passive suspension for high frequency wheel control, +/-
4” @ wheel center

Suspension Tuning
Frequencies

Distribution A, Approved for Public Release 5


Quarter Car Simulation – 63mm RMS Road
Started building a quarter car test rig to prove the concept
Prototype tested at Aberdeen Test Center

Distribution A, Approved for Public Release 6


Full vehicle model

• Full vehicle modeled in ADAMS/Car


– Passive low-travel suspension
– Fully active high-travel suspension
– Driver model native to ADAMS/Car
• Control system modeled in Simulink
– Incorporated in ADAMS model as a discrete Coder dll
– Control parameters passed from ADAMS
– Includes steer-by-wire control
• Hydraulics modeled in Simulink
– Incorporated in ADAMS model as a continuous Coder dll
– Control parameters passed from ADAMS

Distribution A, Approved for Public Release 7


Full vehicle optimization

• Multiple different events setup for the optimization


– To avoid sub-optimal solutions
• 3 different roads at different speeds for absorbed power
• 1 half round for maximum acceleration
• One-sided large obstacle for high-travel stability
• Step steer for roll control and handling stability

Distribution A, Approved for Public Release 8


What can we do?

Distribution A, Approved for Public Release 9


Input parameters
Illustration only

• Passive suspension parameters:


– Springs (front and rear) stiffness
– Fox displacement dependent dampers
• 12 parameters for front and 12 for rear
• Hydraulic system
– Gas volumes and initial pressures front and rear
– 6 parameters
– could easily be expanded to many more but time was limited for valve
selection
• Control system
– 23 parameters
• Total: 53 parameters

Distribution A, Approved for Public Release 10


Outputs

• Objectives
– Minimize absorbed power over three uneven roads
– Minimize vertical acceleration over two half rounds
– Roof obstacle
• Minimize roll and pitch
• Constraints
– Step Steer
• 5% < Step steer roll overshoot < 15%
• 3% < Step steer yaw rate overshoot < 18%
– Roof obstacle
• Don’t let the floor hit the ground

Distribution A, Approved for Public Release 11


Optimization setup

• SHERPA
– Between 150 to 1500 runs
• Many, many optimizations were performed
– Due to
• Control system development
• Hydraulic hardware selection
• Design iterations
– For 6 months, optimizations were performed continuously
• Some produced great progress
• Many could not even get us back to the previous baseline
• The first optimization produced the best results
– But then we started adding in reality (valve control, uncertainty, friction…)

Distribution A, Approved for Public Release 12


2.5” RMS road at 60 mph
8.43 W absorbed power before optimization

4.2 W absorbed power after optimization

Distribution A, Approved for Public Release 13


6W, 12W, 20W Speeds
All roads with 6W speed > 90 mph removed
Driver Absorbed Power
100.0 Easy to get “speed blind”
90.0
Very disappointed when we
80.0
could not reach target speeds
Speeds, mph

70.0

60.0
Have to remind ourselves that
50.0
the targets are 4-5 times
40.0
higher than anything fielded
EVO 3, 6W speed

30.0 EVO 3, 12 W speed


today
EVO 3, 20W speed
20.0
Treshold speed
10.0 Objective speed

0.0

Distribution A, Approved for Public Release 14


So what does all theses numbers mean?
Both these vehicles runs at speed where the absorbed power is 6W

Distribution A, Approved for Public Release 15


• Project Close-out
759 Days

54,962
Hours

Distribution A, Approved for Public Release 16


Q&A

Distribution A, Approved for Public Release 17

You might also like