A terrain treadmill to study small animal locomotion through large obstacles

A major challenge to understanding locomotion in complex 3-D terrain with large obstacles is to create tools for controlled, systematic lab experiments. Existing terrain arenas only allow observations at small spatiotemporal scales (~10 body length, ~10 stride cycles). Here, we create a terrain treadmill to enable high-resolution observations of small animal locomotion through large obstacles over large spatiotemporal scales. An animal moves through modular obstacles on an inner sphere, while a rigidly-attached, concentric, transparent outer sphere rotated with the opposite velocity via closed-loop feedback to keep the animal on top. During sustained locomotion, a discoid cockroach moved through pillar obstacles for 25 minutes (≈ 2500 strides) over 67 m (≈ 1500 body lengths), and was contained within a radius of 4 cm (0.9 body length) for 83% of the duration, even at speeds of up to 10 body length/s. The treadmill enabled observation of diverse locomotor behaviors and quantification of animal-obstacle interaction.


23
In nature, terrestrial animals often move through spatially complex, three-dimensional terrain 1 .

24
Small animals are particularly challenged to traverse many obstacles comparable to or even larger than 25 themselves 2 . By contrast, the majority of laboratory studies of terrestrial locomotion have been performed 26 on flat surfaces 3-10 , either rigid or with various surface properties (friction, slope, solid area fraction, 27 stiffness, damping, ability to deform and flow, etc.). 37 Furthermore, such linear treadmills allow only untethered movement along one direction. Alternatively, 38 spherical treadmills use lightweight spheres of low inertia suspended on air bearing (kugels) to allow small 39 animals to rotate the spheres as they freely change their movement speed and direction, 25,26 . However, the 40 animal is tethered, and obstacles cannot be used.

41
Here, we create a terrain treadmill (Fig. 1A, B) to enable large spatiotemporal scale, high-resolution 42 observations of small animal locomotion in complex terrain with large obstacles. Our terrain treadmill 43 design was inspired by a celestial globe model (Fig. S1). The terrain treadmill consists of a transparent, 44 smooth, hollow, outer sphere rigidly attached to a concentric, solid, inner sphere using a connecting rod 45 (Fig. 1A, Video 1). Terrain modules can be attached to the inner sphere (Fig. 1A, B, C) to simulate obstacles 46 that small animals encounter in natural terrain 16 . The outer sphere is placed on an actuator system 3 consisting of three actuated omni-directional wheels (Fig. 1A). An overhead camera captures videos of the 48 animal moving on top of the inner sphere, with an ArUCo 27 marker attached on its body. The animal's 49 position estimated from tracking the marker is used by a feedback controller (Fig 2A) to actuate the 50 connected spheres with the opposite velocity to keep the animal on top (Fig. 3) as it moves through the 51 obstacle field (Fig. 4, 5, Videos 2, 3). Finally, the reconstructed 3-D motion can be used to estimated 52 different metrics such as body velocities and antennal planar orientation relative to the body heading (

79
The Kalman filter continued to estimate the animal's position even when the marker was obscured from 80 body rolling (Fig. 5A) or the outer sphere's seam (Videos 1, 2). In addition, over the course of 12 trials, the 81 animal freely explored and visited almost the entire obstacle field (Fig. 5G, H). Finally, the animal's motion 82 relative to the treadmill was used to estimate metrics such as body velocity components (

154
Our study is only a first step and the terrain treadmill can use several improvements in the future 155 to realize its potential. First, we will add more cameras from different views to minimize occlusions and 156 diffused lighting from different directions to minimize shadows, as well as increase camera frame rate to 157 accommodate rapid antenna and body movement, to achieve more reliable tracking of the animal body and

171
Both spheres were arranged concentrically using a rigid connecting rod passing through the sphere centers, 172 with a 10 cm space between surfaces of both spheres. To ensure that the connecting rod passed exactly 173 through both sphere centers, we made custom support structures (Fig. S2B, C) to precisely drill through 174 both the inner and outer spheres. The inner sphere was secured to the connecting rod using shaft collars on 175 both sides (Fig. S2A, i). The ends of the connecting rod had threaded holes for the outer hemispheres to be 176 screwed on to it (Fig. S2A, ii). The two outer hemispheres were then mated and sealed using clear tape about an axis that is perpendicular to the motor axis and tangential to the wheel rim (Fig. 2B). We coated 185 the rollers with a layer of protective rubber (Performix Plasti Dip) to reduce their chance of scratching the 186 transparent outer sphere. The three motors were equally spaced around the base (Fig. 2C) and tilted by 45° 12 (Fig. 2D). The tilt angle was chosen based on the size of the base to allow each omni-directional wheel to 188 be perpendicular to the sphere at the point of contact (Fig. 2D), which reduces vibration and simplifies 189 actuation kinematics. Each DC motor also had an encoder to measure and control its rotation speed and was 190 powered from a 12 V DC power supply. To measure the animal's movement relative to the pillar obstacle field, we first measured the 286 movement of the pillar obstacle field (i.e., treadmill rotation) relative to the camera. We attached 31 ArUCo 287 markers to the inner sphere, with one each at the center of hexagonal and pentagonal regions of the soccer 288 ball pattern projected on the sphere (Fig. S4). We then separately created a map of all markers attached on 289 the inner sphere (referred to as marker map, Fig. S4, right) using ArUCo Marker-mapper application.

290
Because each marker and its four corners were fixed relative to the coordinate frame attached to the inner 291 sphere (i.e., T3 is known, Fig. 5D), when one of the markers on sphere is tracked (i.e., T1 can be measured, axis along x axis. Because during portions of a trial the animal body marker was not tracked for a long 307 duration, we did not consider those video frames. As a result, each trial was assumed to be composed of 308 17 multiple segments, and each of their equivalent 2-D trajectories were assumed to have the same initial 309 conditions as described above (Fig. 6D).

310
Maintenance 311 To prevent occlusions and allow reliable camera tracking, the transparent outer sphere must be 312 wiped clean after every use to remove any smudges off the surface. Because wiping with regular cloth 313 towels may scratch the outer sphere, we used a microfiber cloth (AmazonBasics) with soap and water. In