This paper describes a wheeled underwater robot developed for locating chemical sources autonomously under stagnant flow conditions. In still water, the released chemical stays in the immediate vicinity of the source location. The search for chemical sources under such conditions is extremely laborious since the presence of a chemical source cannot be detected from a distant place. The chemical sensors on the robot show no response unless a chemical substance released from the source arrives at the sensors. Crayfish in search of food are known to actively generate water currents by waving their small appendages with a fan-like shape. It is considered that the generated water currents help their olfactory search. The smell of food is carried to their olfactory organs from the surroundings by the generated flow, and then is perceived. The robot presented in this paper employs arms mimicking the maxillipeds of a crayfish to generate water currents and to draw chemicals to its sensors. By waving the arms vertically, a three-dimensional flow field is generated and water samples are drawn from a wide angular range. The direction of a chemical source can be determined by comparing the responses of four laterally aligned electrochemical sensors. Experimental results show that the flow field generated by the maxilliped arms is more effective in collecting chemical samples onto the sensors than that generated by a pump. The robot equipped with the maxilliped arms can detect the presence of a chemical source even if the source is placed off the trajectory of the robot.
Many aquatic animals rely on their olfaction when searching for food [1-3]. For example, sharks are famous for their keen sense of smell, and are known to find food by tracking odor plumes. When they perceive a smell of food, they proceed in the upstream direction [
Olfactory signals spread in the environment by diffusion and advection of odor molecules [
Nevertheless, crayfish prefer to live in still water, e.g., at the bottom of a lake or a pond. Even though crayfish are known to show upstream plume tracking behavior in strong water flow [
The goal of this research project is to develop an underwater robot that can autonomously locate chemical sources. Toward successful applications of such robots in real environments, various technological challenges need to be overcome. They include the development of autonomous underwater vehicle platforms and appropriate chemical sensors, as well as devising effective search strategies and optimizing the sensor configurations. Our focus is on the latter two issues. Two underwater robots with chemical plume tracking capabilities have been reported so far in the literature. Both of them are based on the rheotactic strategies, assuming the existence of sufficiently strong water flow and chemical plumes with well-defined shapes [11,12]. In contrast, here we report a crayfish robot designed to search for chemical sources under stagnant flow conditions. The robot is not only equipped with an array of chemical sensors, but also with
a flow generator to enhance chemical reception by drawing surrounding water samples to the sensors. In our previous work, the underwater robotic system with a suction pump was developed [
The structure of the rest of the paper is as follows. In Section 2, descriptions of the crayfish robot and its prototypes are provided. In Section 3, experimental results on testing different ways of waving the maxilliped arms are presented. In Section 4, comparison is made between the flow fields generated by the maxilliped arms and a pump. In Sections 5 and 6, results of experiments on chemical source localization are summarized. Section 7 concludes the paper.
The detailed analysis of the flow patterns generated by crayfish revealed their high efficiency in collecting odor samples [
As shown in
step motor (SPG20-298, Copal Electronics Corp.) placed above the water surface. The maxilliped arms are waved at 5 Hz by adjusting the speeds of the step motors using a microcontroller (PIC12F683, Microchip). This frequency was chosen to be the same as the frequency at which crayfish wave their maxillipeds [
The body of the robot was made using a waterproof plastic container. A silicone rubber seal placed between the upper rim of the container body and its detachable lid prevents water from seeping in. Commercially available rotary shaft seals (Turcon Roto Variseal, Trelleborg Sealing Solutions) keep water from entering through gaps between the shafts of the driving wheels and the sockets. A weight was loaded to prevent the robot from floating up during the experiments.
The robot is equipped with two independent wheels driven by DC geared motors (TG47C VM-230-KBED, Tsukasa Electric Co.). An EyeBot (JOKER Robotics) is employed as a main controller of the robot. A Motorola 68332 microcontroller operated at 25 MHz processes the received sensor responses and sends the control signals to the motors. The feedback signals from the rotary encoders on the DC motors are used for odometry. The sensor responses and the odometry data are sent to an external PC through serial communication in order to record the data for detailed off-line analysis. A lithiumion rechargeable battery with a capacity of 1500 mAh (NP-400, Konica Minolta Holdings, Inc.) can supply electric power to the robot up to 90 min.
Amperometric electrochemical sensors are used to detect chemical substances. Four carbon working electrodes with a diameter of 0.9 mm are placed at the front part of the robot head. These working electrodes share a silver reference electrode and a stainless-steel counter electrode, which are placed on the bottom of the robot head. A potentiostat circuit controls the voltage between the working electrodes and the reference electrode at a certain set point (0.7 - 0.8 V). The current generated by oxidation or reduction of a target chemical at each working electrode is converted to a voltage output. A dedicated microcontroller (PIC16F690, Microchip) measures the four voltage outputs of the potentiostat circuit, and sends the voltage values to the EyeBot controller. The sensing electrodes are numbered from the left to the right of the robot as shown in
When in search of a chemical source, the robot examines the existence of a chemical substance in the collected water samples, and starts to move if the response of any of the four sensors exceeds a predefined threshold. The robot proceeds in the direction of the sensor with the largest response. When the largest signal is obtained from sensor 1, the robot turns counterclockwise by 8˚ and then moves forward by 6 mm. When the output of sensor 2 is the largest, the turning angle is reduced to 4˚. If sensor 3 or 4 shows the highest response, the robot moves in a similar way except that the turn is made in the clockwise direction. It was sometimes observed that a water sample drawn to the sensor stayed on the carbon working electrode even after the forward movement of the robot. To wait for the water samples around the sensors to be replaced, the robot pauses for one second after each forward movement. The direction determination is performed based on the sensor output values measured after the pause.
Before the fabrication of the crayfish robot described in the previous section, the chemical sample collection efficiency of the maxilliped arms was investigated using prototype chemical sensing systems shown in
To use a pump is probably the simplest and most widely used way for robots to generate water currents. To show the advantage of the maxilliped arms in the effectiveness on collecting water samples on the sensors, the water currents generated by a pump were compared in terms of chemical reception at the sensors with those generated by the maxilliped arms. As shown in
In the experiments on chemical source localization, the robot tried to detect a chemical substance in a small water pool (1300 mm × 800 mm) shown in
Since a certain amount of supporting electrolyte is required for proper operation of the electrochemical sensors, salt (sodium chloride) was added to the background water and the ascorbic acid solution. The concentration of salt was adjusted to be in the range of 0.1 M to 0.485 M. The concentration of 0.485 M corresponds to the salinity of seawater, and was chosen in view of future applications of the robot in marine environments. The water pool holds 104 liters of water, and 2.9 kg of salt is required to attain the same salinity as the seawater. For the repeated experiments in the water pool, the concentration of salt was decreased to 0.1 M to reduce the consumption of salt. Cyclic voltammograms measured in 1 mM ascorbic acid solution with various salinities show that the decrease in the salt concentration to 0.1 M has no effect on the response of the carbon electrode sensor to ascorbic acid.