Locomotion studies and modeling of the long-tailed lizard Takydromus sexlineatus Konstantinos Karakasiliotis*, Kristiaan D’Aoˆ ut, Peter Aerts and Auke Jan Ijspeert, Member, IEEE Abstract— Morphology is an important factor in locomotion. It may guide the control strategies that an animal or a robot uses for efficient locomotion. In this paper we try to understand the locomotion strategies of a lizard with a distinctive feature, the long-tailed lizard Takydromus sexlineatus. We recorded the performance of real animals in terms of forward speed and then developed a simulation model respecting the morphometric characteristics of long-tailed lizards. We then run systematic tests altering several control parameters of the model. The simulation experiments suggested possible control strategies for effective locomotion given this type of morphology. The experiments were not constrained or guided by any prior knowledge on specific animal angular kinematics. Therefore, the good match between the suggested kinematics for optimal speed and the kinematics of the real animal suggests that our framework is capable of exploring in the future the effects of morphosis on the locomotion strategies of animals, i.e. to perform the same study with shorter or no tail. I. INTRODUCTION When it comes to fast, stable and adaptive locomotion, lizards are one of the best animal groups to study. Moreover, lizards display a wide range of morphological diversity and ecological adaptations, including the ability to loco- mote on a variety of substrates ([1]). Lizard locomotor mechanics are remarkably similar to those of other legged animals ([2]) which suggests that similar locomotor strategies might be shared with other tetrapod groups. Understanding how specific morphological variations affect the locomotion strategies of animals may reveal the principles that connect morphology and control. Within the lizard taxon (Lacertilia), in the family of Lacertidae all members are relatively closely related. This increases the chances that observed morphological diversity within this family reflects functional diversity, and not phy- logenetic diversity. Within the Lacertidae, two species, the Lacerta vivipara and Takydromus sexlineatus have similar body size and they display a general “lizard” body shape, i.e. with not extremely strong developed (as is specialist runners) or underdeveloped (as in scincids) limbs. The specialist, T. sexlineatus, differs from the generalist L. vivipara mainly in one, clear, dis- tinguishing feature: extreme tail elongation ([3] – [5]). This should facilitate interpretation of biomechanical comparisons between these species. Lacerta is the generalist reference, and a considerable amount of literature is available on general characteristics of lizard locomotion ([2], [6], [7]). Konstantinos Karakasiliotis and Auke Jan Ijspeert are with the ´ Ecole Polytechnique F´ ed´ erale de Lausanne (EPFL), Switzerland. Kristiaan D’Aoˆ ut and Peter Aerts are with the Laboratory for Functional Morphology, Department of Biology, University of Antwerp, Belgium. Fig. 1. Snapshot of the long-tailed lizard during the experiments. Therefore, the main focus of our study is on Takydromus (Fig. 1) because it sports a clear case (tail elongation) of long-term morphosis. In particular, we address two basic questions: i) What are the control parameters for which the model closely replicates the basic kinematics of the long- tailed lizard? and ii) What body postures can optimize the model’s performance in terms of speed and how do they compare to the real animal? It is important here to note that these questions were not guided by kinematic recordings from the specific animal and, therefore, the optimal solutions from our systematic exploration of control parameters was only dependent on the specific morphological and inertial properties of the model. II. LONG–TAILED LIZARD KINEMATICS A. Materials and Methods For the animal experiments, 15 Takydromus sexlineatus were acquired commercially and housed in a terrarium with plenty of food and water, and the possibility to thermoreg- ulate. Average body mass was 3.80 ± 0.73 g (range: 2.47 – 5.12 g), average total body length was 304 ± 32 mm (range: 225 - 358 mm), and the percentage of total body length that is tail was 82 ± 2% (range: 75 – 85%), i.e. the tail is 3 to 5.6 times longer than snout-vent length. Animals were filmed dorsally using a high-speed digital video camera (300 fps, Casio EX-F1). Subsequently, anatom- ical landmarks were digitized manually, frame-by-frame, and used for conventional gait assessment (spatiotemporal gait variables, speed and frequency). After kinematic experiments, one individual was sacri- ficed, frozen in a straight position and segmented using a sharp scalpel. Frozen body segments were weighed on a Mettler microbalance. The following segments were mea- sured: head, neck, trunk (in five segments of equal length), tail (in ten segments of equal length), upper leg, lower leg and foot (Fig. 2). For each segment, mass, length, and average width (for trunk and tail segments) was measured. These measurements were used then to define the inertial properties of the simulated model (see section III-A and