Design of an Autonomous Jumping Microrobot Sarah Bergbreiter and Kristofer S. J. Pister Berkeley Sensor and Actuator Center 497 Cory Hall Berkeley, CA 94720, USA {sbergbre, pister}@eecs.berkeley.edu Abstract – This paper presents the design and initial results for an autonomous jumping microrobot. At the millimeter size scale, jumping can offer numerous advantages for efficient locomotion, including dealing with obstacles and potentially even latching onto other larger mobile hosts. Robot design is divided into four primary areas: energy storage, actuation, power, and control. Like its biological inspiration, the flea, a jumping microrobot requires an energy storage system to store energy and release it quickly to jump. Silicone micro rubber bands have been fabricated and assembled into the microrobot for this task. To stretch these micro rubber bands, electrostatic inchworm motors are chosen as actuators due to their high forces, long throw, and low input power requirements. Finally, solar cells and a microcontroller have been chosen to power and control the microrobot. A small-scale version of this system has been prototyped with the solar cells and a simple 4-bit microcontroller driving an inchworm motor. Separately, an inchworm motor has been demonstrated pulling and storing 4.9 nJ of energy in a micro rubber band. Finally, initial tests with a probe-loaded robot prototype have demonstrated a microrobot which can potentially jump 1.2 cm straight up. Index Terms – Micro/Nano Robots, Jumping Robots, Biologically-Inspired Robots. I. INTRODUCTION Mobile autonomous microrobots, defined as millimeter-sized mobile robots with power and control on board, offer numerous advantages due to their size and low power requirements. Microrobots at this size scale could be used to add mobility to sensors in large-scale sensor networks as the size of those integrated sensors shrink as shown in [1]. Large numbers of autonomous mobile microrobots could also be used for search in unstructured environments, surveillance, and micro construction tasks. However, obstacles present a serious challenge to mobility at the millimeter-scale due to the fact that even surface roughness can become an issue for movement. Moving around in an unstructured environment becomes even more difficult. Flying microrobots, such as the one outlined in [2], overcome this predicament by simply ignoring terrestrial considerations. However, such robots can be difficult to control and to design autonomously due to the continuous high power output required from the actuators. Walking microrobots as seen in [3] offer a much simpler design and control problem, but can only overcome obstacles on the same order of magnitude as their leg length. Efficiency becomes another important issue for mobile robots at the millimeter scale. When the power supply becomes a significant portion of the robot’s mass, it is essential to design the robot to move as efficiently as possible. Efficiency can be improved by choosing an appropriate gait and reducing the energy required to move internal pieces of the robot which don’t contribute to external motion. In nature, millimeter-sized insects often address both obstacles and efficiency through jumping. While jumping is often seen as an energetically costly escape mechanism, [4] has shown that as insect size shrinks, jumping becomes more advantageous due to the higher takeoff velocities allowed. Because small jumpers are more mechanically efficient than their larger counterparts, they require less muscle tissue to make them more energy efficient as well. As an example of a jumping insect similarly sized to the proposed robot, the froghopper has a mass of approximately 12 mg and averages 43 cm vertical per jump with a 58o takeoff angle [5]. For the microrobot to gain these same advantages, a jumping gait is proposed. In this paper, jumping is defined as ballistic jumping much like a frog or flea. Continuous jumping, or hopping, requires significant control challenges that are not addressed in the current work. Jumping robots have been demonstrated previously at larger sizes. In [6], a 1.3 kg jumping robot was designed as a potential platform for planetary exploration. Jumps of up to 3 m horizontally and 1.2 m vertically were demonstrated with a single actuator to compress a spring and right itself after a jump. A prototype 10 g robot was designed and simulated in [7] to explore jumping as an option for locomotion at centimeter size scales. This paper presents the design, fabrication, and some initial results for an autonomous jumping microrobot. Section II will discuss the challenges for jumping at this scale as well as requirements for energy storage, actuation, power, and control. Section III details the fabrication process used to build this microrobot and Section IV discusses some initial results from prototypes. II. MICROROBOT DESIGN The ultimate goal of this project is to create an autonomous mobile microrobot that can move around in unstructured environments. Ideally, this microrobot will be millimeter-sized and be able to jump many centimeters several times per minute. To examine the initial kinetic energies required to produce such jumps, a simple model describing the robot as a point mass of 10 mg (similar mass to [3]) is used to plot various trajectories in Figure 1. Given a specified kinetic energy at the jump take-off, height and distance may be calculated as follows.mgUdkinetic/2sin2!= ( 1 ) mgUhkinetic/)(sin2!= ( 2 ) In (1), d is horizontal distance traveled, Ukinetic is the kinetic energy available at take-off, ! is take-off angle, h is maximum height, m is robot mass, and g is gravity. As can be seen in Figure 1, kinetic energies as small as 1 µJ will still lead to 1.8 cm of horizontal travel over millimeter-sized obstacles and energies of 25 µJ will lead to jumps greater than 40 cm overcoming obstacles over 10 cm high. However, the height and distance numbers given by (1, 2)