Engineers and scientists have continuously pushed the field of robots and their development of technologies that would allow the creation of true artificial life. As part of their effort, much of their current research surrounds the field of energy autonomy, or rather, developing a robot that can locate, acquire, and use energy from its surroundings through a completely self-directed process. This paper presents the evolution of autonomous robots created at the Intelligent Autonomous Systems Laboratory (IASL) at the University of Bristol.
The EcoBot was motivated by the idea that future robots will not only need to be able to acquire energy from their environment in order to perform standard tasks, they will need to manage their energy supply and decide between activity and inactivity accordingly. [1] Early research in the subject lead to the creation of the SlugBot, a robot that could essentially detect and collect slugs to essentially use as a bio-fuel. The robot would ferment the slugs to product biogas that would pass through a methane fuel cell to produce electricity that could be stored in on-board batteries. [2]
The EcoBot I, represented in Fig. 1 below, is the 0.96 kg robot that followed SlugBot and focused primarily on recreating digestive processes in nature in which the robot would "breathe" and "eat" to break down oxygen and food components into produce usable forms of power. The 22 cm diameter robot found and acquired energy sources through use of a photo-tactic (light seeking) process, in which threshold voltages were measured and directed the robots motion in bursts. The final system moved at a speed of 2.4 meters per hour.
|
|
| Fig. 1: Schematic Diagram of the EcoBot I. University of Bristol, Intelligent Autonomous Systems Laboratory, EcoBot Team. (Courtesy of IAS EcoBot Team.) |
To convert energy supplies into useful forms, the robot used microbial fuel cell (MFC) technology, in which microbes inside a semi-permeable container would be used to extract electrons from a passing nutrient and store those through use of an electrode, using E. coli with refined sugar as a catalyst and essentially creating an artificial metabolism. The energy extracted from this process is then stored in a bank of capacitors. The supply is tracked via photo-diodes that set off two high efficiency motors when the energy storage drops below a certain level. [2] In its final version, the charge/discharge cycle of the EcoBot was 30 seconds and 2 seconds respectively, a 99% improvement (reduction) versus a comparable autonomous robot developed at the University of South Florida by Stuart Wilkinson.
The EcoBot II, the IASL’s next project, was created to acquire its energy from physical food sludge, such as rotten apples and flies, as opposed to the refined sugars collected by the EcoBot I. Depicted in Fig. 2 below, the device was designed to wirelessly transmit the current value of air temperature at various locations in a playing field, collecting and storing its own energy along the way in order to do so. [3] In addition, they added an oxygen gas diffusion cathode, allowing the robot to utilize ambient air for additional energy. The raw substrate from the sludge serves as the biological catalysts at the anode while the oxygen from the air serves as the oxidizing agent to move the electrons and produce water, closing the circuit at the cathode.
 |
| Fig. 2: Schematic Diagram of the EcoBot II. University of Bristol, Intelligent Autonomous Systems Laboratory, EcoBot Team. (Courtesy of IAS EcoBot Team.) |
The resulting movement of the robot is pulsed, dependent upon the current amount of banked energy in the capacitors. When accumulated energy drops below a certain level, the robot becomes inactive until the MFC can produce and store enough energy, at which point motion is resumed after passing a higher set value.
Compared to an alkaline battery, the EcoBot II has plenty of room for improvement. [3] The robot's MFC, which costs about $5, has an energy content of 37 milliwatt-hours (133 joules) and a mass of 0.22 lbs (0.1 kg), giving an energy density of 1,330 J/kg. A typical AA alkaline battery, by contrast, has an energy content of 4.2 watt-hours (15,120 joules) and a mass of 25 g (0.025 kg), giving an energy density of 600,000 J/kg.
Autonomous robots pose a variety of interesting challenges to engineers, chemists, and biologists working on the acquisition and storage of useful energy. While the EcoBot's MFC metabolism certainly enables the robot to carry out its essential tasks (there are no other forms of stored energy on board), the robot on which the system has been tested is small and requires a small of amount of power to perform its simple tasks.
The important figure here is the energy density of the MFC, a mere 1,330 J/kg compared to more than 600,000 J/kg for a standard AA battery. In essence, any MFC would have to be about 450 times as massive as a AA battery in order to output the same amount of energy. Because this is obviously not feasible, any complicated task performed by most modern robots would require a much stronger fuel cell with the capacity to store far more power than the MFC. In addition, the robot's ability to make decisions based on levels of stored energy would need to be far less discontinuous, as the EcoBot currently switches between modes of complete activity and inactivity. True artificial life would require more complex artificial intelligence to smooth the curve and gradually reduce activity as it approaches a certain threshold.
© Fidel Hernandez. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
[1] I. Kelly, "The Design of a Robotic Predator: The Slugbot," Robotica 21, 399 (2003).
[2] I. Ieropoulos, C. Melhuish, and J. Greenman, "Artificial Metabolism: Towards True Energetic Autonomy in Artificial Life," Adv. in Artificial Life 2801 792 (2003).
[3] I. Ieropoulos et al., "EcoBot-II: An Artificial Agent With a Natural Metabolism," Int. J. Advanced Robotic Systems 2, 295 (2005).