Head and leg adjustments help insects and legged robots traverse cluttered, large obstacles

Department of Mechanical Engineering, Johns Hopkins University ywang460@jh.edu https://li.me.jhu.edu

Wang, Othayoth, Li, J. Exp. Biol., in prep.

1. Background & Motivation

Animals integrate active sensory feedback and passive mechanical feedback to control locomotion

Potential energy landscape

Physical interaction (energy landscape) plays a major role during obstacle traversal

Model system Locomotor transition pathways

Transition barrier

Othayoth et al. (in press), PNAS

Dickinson et al. (2000), Science

Saddle

Awareness of landscapehelps find a lower barrier

Hypothesis: Cockroaches actively adjust body and appendages to overcome barrier and make transition

2. Experimental Observations

Leg sprawl

Head Bending

Pushing Leg Height

Supporting Leg Height

0

5

10

15

0

5

10

15

0

50

100

Leg sprawl angle (⁰) s.d. of head flexion (⁰)Leg height difference (mm)

Difference

*** *** ******

Frequent head flexionLeg sprawl Differential leg use

Run Push Roll

Beam obstacle traversal can be divided into three phases

Cockroaches actively adjust their head and legs to make locomotor transition

Beam obstacle layer

Cameras(100 fps)

3-D kinematics tracking setup

Reduces roll stability Generates roll torque Sense terrain (?)

Run Push Roll

0.30s 0.43s 1.06s 1.29s 1.97s 3.13s 3.44s

Tags

Beads

Beads

N = 8 animals, n = 64 trials

P < 0.05 ANOVA

P < 0.05 ANOVA

P < 0.05 ANOVA

Run Push RollRun Push Roll

Pitch

Saddle

Body roll (°)

`Body pitch (°)

`

>0.54

Energy(mJ)

50

-50

Roll

-50

-50

0

0.4

0.52

0.462

0

3. Energy Landscape ModelingEnergy landscape approach to understand physics of locomotor transitions

Proper head adjustment reduces energy barrier on landscape

Body pitch (º)

−80

−40

0

40−70

0

70

Body roll (º)

0

25

0

50Start

1

Start

Transition barriers

Roll

Start

1

Roll

Pitch Pitch

3

Energy (mJ)

iiiiii

Beams

Start

Pitch

Pitch

Start Start

Roll

Vertical vibration

3−80º

0

40−70

0

70

3

3

1

Start

i ii iii iii’

17

7

Energy (mJ)

11

8

Energy (mJ)

0

20

2

−80

−40

0

40−70

0

70

Barrier without head adjustment

Pitch

Roll

Saddle

Energy barrier comparison

Othayoth et al. (in press), PNAS

0

300

600

−40 −20 0 20 40

Traversalbarrier

(µJ)

x (mm)

Roll more favorable

K = 11.44 mN·m/rad

− Roll− Pitch

Body roll (º)

−90

0

9090

0 −90

Body pitch (º)

Potential energy (µJ)2.5

×104

2

1.5

1

0.5

With head bending(measured head angle)

Representative barrier profileHead adjustment lowers barrier

3-D reconstruction

Pitch

Saddle

Body roll (°)

`Body pitch (°)

`

>0.54

Energy(mJ)

50

-50

Roll

-50

-50

0

0.4

0.52

0.462

0

Without head bending(constant head angle)

Body roll (°)0 20 40 60 80

Barrier with head adjustment

Energy(mJ)

0.38

0.5

0.4

0.42

0.44

0.46

0.48

0 0.5 1 1.5 2 2.5Time (s)

0

0.1

0.2

0.3

0.4

No head bendingWith head bending

Barrier energy(mJ)

Start rolling Roll to max

Time (s)0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0 10 20 30 40 50 60

Trial NO.

-0.08

-0.06

-0.04

-0.02

0

0.02 Barrier is smaller with head bending than without

(mean change < 0)

(P < 0.05, ANOVA)

0.02

0

-0.02

-0.04

-0.06

-0.08

Change in energy barrier(mJ)

0 10 20 30 40 50 60

Trial No.

4. Robophysical Model & Future Work

Next steps

Beams

Retract Left Leg

Extend Right Leg

Head bending

Sprawlinward

Gap = 57% Body WidthInitial Pitch = 45°

Legged robot to systematically control and vary active adjustment of head and legsHead flexionLeg sprawl Differential leg useLegged robot

Preliminary result: all the three adjustments help the robot roll into the gap

Low cost 3-axis force sensor(designed by J. Krakauer’s group)

Y

Z

X

Embedded sensors at the neckRobot with force sensors

Force measurements to understand:• Relationship between contact forces and

potential energy landscape (slope?)• Transitions on landscape emerging from local

force sensing

Acknowledgements:Yuanfeng Han and Qiyuan Fu for help with imaging setup;Yuanfeng Han, Qiyuan Fu, QihanXuan and Xiao Yu for discussion;Xiao Yu for help with animal care.

5 trials5 trials5 trials

11

Yaqing Wang, Ratan Othayoth, Chen Li