Adam J. Rutkowski

Moth-Inspired Plume Tracking Strategies In Three-Dimensions

Two simple three-dimensional moth inspired odor tracking algorithms, Counter-turner and Modified counter-turner, were tested on a robotic platform. The Counter-turner uses the plume edge to modify the timing of the crosswind movements, while the Modified counter-turner uses the plume centerline. Both algorithms shows some success in tracking the plume to it’s source. In addition, flight tracks show promise in mimicking the flight tracks observed in biological experiments with the moth Manduca. Sexta.

A robotic platform for testing moth-inspired plume tracking strategies

A mobile autonomous robot capable of tracking an odor plume to its source can be used to locate hazardous material spills or leaks. To test plume tracking strategies in a laboratory environment, a robotic platform consisting of a linear Cartesian robotic gantry mounted inside a wind tunnel with a mobile floor has been designed. A plume of ionized air is created by an ion detector installed in the wind tunnel. A two-dimensional plume tracking strategy based on the behavior of the tobacco hornworm moth Manduca sexta is implemented and tested. It is discovered that a plume tracking algorithm can be implemented with the robotic platform almost as easily as in simulation.

Three-dimensional characterization of the wind-borne pheromone tracking behavior of male hawkmoths, Manduca sexta

We studied the relationship between vertical and lateral movements during free flight odor plume tracking by male moths, Manduca sexta, in a wind tunnel with the “horizon” set at different altitudes. Three-dimensional recordings revealed that the plume tracking males generated roughly equivalent movements vertically and laterally regardless of horizon height. We hypothesized that the moths’ tracks would be narrower in the vertical plane when they were presented with visual patterns on the tunnel’s side walls. Instead, we discovered that their tracks tended to be wider in the horizontal plane. Anecdotal observations of other moth species describe plume tracking flight in three dimensions as “spiraling”, suggesting a specific predictable relationship between vertical and lateral movements. However, we found that the relative phase, frequency, and amplitude of the vertical versus lateral movements vary on a maneuver-by-maneuver basis with no predictable temporal or spatial relationship. Our analyses suggest that a moth’s trajectory in 3D can best be described as progressing upwind toward the source while cutting through the plume from all directions with loops of different radii. This is a more precise description than the terms “zigzagging” and “counter-turning” which were derived from 2D analyses of this behavior.

Fusion of inertial, optical flow, and airspeed measurements for UAV navigation in GPS-denied environments

This paper describes the data fusion approach that is developed for navigation of autonomous unmanned aerial vehicles (UAVs) for those applications where the Global Positioning System (GPS) signals are denied. Example scenarios include navigation under interference and jamming and urban navigation missions. The system architecture is biologically inspired and exploits measurements that are utilized by flying insects for self-localization purposes. The data fusion algorithm implements the Kalman filter mechanization that fuses INS data (position velocity and attitude), optical flow data from a monocular downward looking visual system (scaled body-frame vehicle velocity components), and compass measurements (azimuth angle). Kalman filter measurement observables are formulated in a complimentary form, i.e., as differences between optical flow/compass measurements and INS states that are projected into the measurement domain. The filter estimates inertial error states and error in the flight height. We present the navigation solution architecture and demonstrate its feasibility using simulations and actual data experiments. Also, we compare our results to a data fusion algorithm that fuses airspeed and optical flow measurements.

Simulated odor tracking in a plane normal to the wind direction

A new strategy is developed for tracking an odor plume in an environment where wind is present. The strategy is inspired by the mechanisms that animals use to orient to stimuli. A male moth counter-turns horizontally and vertically across the wind while tracking a pheromone plume. The inter-turn duration, or the amount of time between turns, is about 500 milliseconds on average in both directions. This observation has lead to the development of odor tracking algorithms that use inter-turn timers to control when to turn in both directions. The algorithm presented in this work takes a different approach. The new algorithm controls the rate of turning of the tracking vehicle in a plane normal to the wind direction. Simulation results demonstrate that counter-turning can be achieved in both directions without the use of inter-turn timers. Also, the estimated odor source location approaches the true source location as time progresses

Egomotion estimation with optic flow and air velocity sensors

We develop a method that allows a flyer to estimate its own motion (egomotion), the wind velocity, ground slope, and flight height using only inputs from onboard optic flow and air velocity sensors. Our artificial algorithm demonstrates how it could be possible for flying insects to determine their absolute egomotion using their available sensors, namely their eyes and wind sensitive hairs and antennae. Although many behaviors can be performed by only knowing the direction of travel, behavioral experiments indicate that odor tracking insects are able to estimate the wind direction and control their absolute egomotion (i.e., groundspeed). The egomotion estimation method that we have developed, which we call the opto-aeronautic algorithm, is tested in a variety of wind and ground slope conditions using a video recorded flight of a moth tracking a pheromone plume. Over all test cases that we examined, the algorithm achieved a mean absolute error in height of 7% or less. Furthermore, our algorithm is suitable for the navigation of aerial vehicles in environments where signals from the Global Positioning System are unavailable.

Biologically inspired self-motion estimation using the fusion of airspeed and optical flow

The problem of determining the self-motion of an autonomous aerial vehicle using onboard sensors is of interest to both engineers and biologists. Engineers typically use some combination of a global positioning system receiver, inertial measurement units, and optical sensors to estimate self motion. Biological systems also possess visual and inertial sensors. Airspeed sensors are also available to both engineering and biological system. The method of self-motion estimation proposed in this paper is to fuse optical flow with airspeed information. Simulation results indicate that this approach can be used to calculate ground speed and wind speed while moving

A Sensor Fusion Approach to Odor Source Localization Inspired by the Pheromone Tracking Behavior of Moths

An approach to odor source localization with an aerial vehicle using the fusion of odor sensors, visual sensors, and airspeed sensors is presented. The motion of the tracking vehicle is decomposed into two components - a component normal to the wind direction and a component tangential to the wind direction. The tangential component is controlled with a strategy that moves upwind when odor is detected and moves gradually downwind when odor is lost. The normal component of velocity is controlled by two different algorithms. The first algorithm controls the rate of turning in the plane normal to the wind direction as a function of concentration. The second algorithm controls the rate of turning and the magnitude of the normal component of velocity as a function of the time derivative of concentration. Both algorithms display the potential to declare the location of an odor source in a three-dimensional space.

Path planning for optimal cooperative navigation

For many navigation scenarios, it is known that the accuracy of navigation state estimates depends on the path traveled, particularly when access to external navigation aids such as the Global Positioning System (GPS) is not available. In this work, we present a path planning method that attempts to minimize the navigation uncertainty of a pair of autonomous vehicles traveling from known initial locations to desired goal locations. For the scenario considered in this study, each vehicle has an onboard odometer for measuring relative changes in position and heading. The vehicles also have sensors for measuring the range between the vehicles. Navigation state estimates are obtained using a centralized batch factor graph approach implemented with the Georgia Tech Smoothing and Mapping (GTSAM) library. In previous work on this problem, candidate vehicle trajectories were chosen from a restricted class of trajectories referred to as pseudo-zigzagging. An exhaustive search of all possible trajectories in this class revealed that the final position uncertainty could be reduced by a factor of 5 when compared with the case of each vehicle traveling straight to its goal location. In the work we present here, the trajectories are no longer restricted to such a limited class. Instead, each path is constructed from a small set of waypoints that may be placed anywhere. We have devised a method that takes an arbitrary set of waypoint locations and adjusts their locations so that the resulting path meets travel time and maneuverability constraints. Using just 3 waypoints for each vehicle path, and applying a simple random search optimization algorithm to the waypoint locations, the final position uncertainty can be reduced by another factor of 3. Thus, we achieve a total factor of 15 reduction in position uncertainty when compared to the straight path (i.e. worst) case. Also, the results indicate that the final position uncertainty can be different for each vehicle. Furthermore, the navigation certainty does not necessarily improve with increased travel time. Thus, travel time should be considered a free parameter (i.e. the time constraint is not necessarily an active constraint). As is the case for most practical nonlinear optimization problems, there is no guarantee that the results obtained in this work are globally optimal. Nevertheless, it is quite clear that path planning can be used to significantly reduce navigation uncertainty for the scenario under consideration.