R

Guidance

10.0 m

Table 4. Parameter settings for CPT strategy for the April 2003 SCI and June 2003 Duck experiments.

Table 4. Parameter settings for CPT strategy for the April 2003 SCI and June 2003 Duck experiments.

Two types of missions were of interest during this set of experiments. The first mission type, labeled ST, contained a single chemical source in the OpArea. The ST mission was intended to find the plume, to trace a plume over a long distance, and to declare the source location. This mission demonstrates detection and tracing of plumes over long distances. The second mission type, labeled MT, may contain a few chemical sources in the OpArea. In an MT mission, the OpArea will be divided into subregions. The AUV will search each subregion for chemical until one of three events occurs. First, the search within a subregion may timeout. In this case, the subregions is declared source free and the AUV moves on to the next subregion. Second, the AUV may detect chemical and declare a source location within the region. It will then move on to the next subregion. Third, the AUV may trace chemical to the upflow edge of the region. In this case, a source will be declared at the intersection of the plume with the upper edge of the subregion and the AUV will move on to the next subregion. When the declared source locations are analyzed at the end of an experiment it is up to the test director to decide whether source locations at the edge of a subregion are due to sources near that location or the result of plumes generated by sources in the adjacent region.

The AUV for these tests was the Albacore REMUS owned by SPAWAR in San Diego, CA. The REMUS was modified to contain a PC104 computer to run the AMP CPT algorithms. The AMP computer received sensor data from the REMUS computer via serial port, processed the sensor data, and output heading, speed, and depth/altitude commands to the REMUS computer via the same serial port. Up and down looking acoustic Doppler current profilers (ADCP) were onboard the REMUS. The AUV also had a CTD mounted onboard, but it was not used due to its slow response time. Also, the AUV used long baseline transducers with acoustic buoys in conjunction with dead-reckoning based on ADCP data to determine onboard AUV position. Finally, a fluorometer was mounted near the nose of the AUV. The fluorometer was capable of detecting Rhodamine dye from a source that was used to create the plume for these experiments. The fluorometer sample rate was 10 Hz. Fig. 7 and Fig. 8 show the trajectory (solid line), chemical detection locations (x's), and declared source location (black dot) for two missions performed at Duck, NC in June of

Fig. 7. Trajectory and chemical detection points. The dashed rectangle is the operating area boundary. The solid curve is the AUV trajectory. Each x marks the location of a chemical detection. The black dot at (N,E) = (414,-242)m marks the declared source location.

2003. The boundary of the OpArea is indicated by the dashed line. The figures use a coordinate system that is defined in the north and east directions relative to the center of the OpArea. These experiments were performed in 4-8 m of water. The bottom was gradually sloping from the coast. The coast is approximately 400 m to the left of boundary of the OpArea in both figures in this section. During all experiments included herein, the water column consisted of a top layer flowing northerly with a speed near 20-25 cm/ s and a bottom layer flowing southerly with a speed near 10 cm/s. The depth of the boundary layer between these two flow regimes changed with location and time.

Fig. 7 shows the trajectory, chemical detection locations, and declared source location for an ST mission. For this mission, the OpArea was 367 x 1094 m (greater than 60 football fields). During this experiment, the flow calculated on the AMP varied in magnitude between 10 and 15 cm/ s and in direction between 110 and 147 deg. For this experiment, the commanded speed was 2 m/s and the commanded altitude was 2 m. Note that the actual altitude varies by plus or minus 0.7 m relative to the commanded altitude. To challenge the CPT algorithm, we wanted the first chemical detection to occur as far as possible from the chemical source. Therefore, the source is located near the upflow edge of the OpArea and the AUV starts the mission near the downflow edge of the box. The AMP CPT algorithms start as soon as chemical is detected. This mission tracks the chemical plume for 976m between the first detection point and the declared source location. The source is declared at 36n11.028, 75w44.620. The ground truth source location is 36n11.035, 75w44.621 as found from sidescan data acquired during a post-declaration maneuver centered on the declared source location. The declared source location is 13 m south and 2 m east of the sidescan sonar location. Note that this error is predominantly in the direction of the flow, as expected. Fig. 8 shows the trajectory, chemical detection locations, and declared source locations for an MT mission. The four subregions are outlined by dashed lines in Fig. 8. During this experiment, the flow calculated on the AMP varied in magnitude between 20 and 30 cm/s and in direction between 160 and 175 degree. For this experiment, the commanded speed was 2 m/ s and the commanded altitude was 1.5 m. The southwest region is explored first. Chemical is detected and tracked for 351 m to the boundary between the southwest and northwest regions. The source for the first region is declared (correctly) at this boundary. Then, AMP drives the AUV to the northwest region. In the northwest region, the plume is tracked for an additional 180 m with a source declared at 36n11.034, 75w44.621. Sidescan sonar data confirmed the source at 36n11.037 , 75w44.622. The error between these locations is 6 m in the downflow direction. Note that this declared source is the same as that (for the same quadrant) from the missions shown in Fig. 7. Note that the latitude and longitude of the declared and sonar source locations match closely between these figures. After declaring the source in the northwest region, AMP drove the AUV to the southeast region and restarted the CPT algorithm. During the transition from the northwest region to the southeast region using the Go To command, chemical detections are ignored. In the southwest region, chemical is detected and tracked a distance of 351 m to a source that is declared (correctly) on the boundary between the southeast and northeast regions. Then AMP drives the AUV to the northeast region. In the northeast region, the plume is tracked for an additional 185 m with the source declared at 36n11.079, 75w44.468. Sidescan sonar data confirmed the source at 36n11.087, 75w44.450. The error between these locations is 31 m in the crossflow direction. This crossflow error is clearly visible in the northeast region of Fig. 8. This crossflow error is an artifact of a navigation fix that occurred prior to the declaration and the declaration logic that only accounted for position differences in the direction of the flow. This will be fixed in future versions of the algorithm.

Fig. 8. Trajectory and chemical detection points. The dashed rectangle is the operating area. The solid curve is the AUV trajectory. Each x marks the location of a chemical detection. The black dots mark the declared source location.

Fig. 8. Trajectory and chemical detection points. The dashed rectangle is the operating area. The solid curve is the AUV trajectory. Each x marks the location of a chemical detection. The black dots mark the declared source location.

Note that in spite of the exact same strategy and parameters being used in all runs, the nature of the trajectories shown in Fig. 7 and 8 during the plume tracing phase look different. Therefore, the differences in experimental conditions deserve comment. First, the mission shown in Fig. 7 was one of the first trials at Duck NC. Due to the fact that we were operating in an unknown environment, the commanded altitude for that mission was 2.0 m. For the mission corresponding to Fig. 8, the commanded altitude was 1.5 m. Analysis of the log files show that plume tracing for the mission of Fig. 7 frequently used the Track-Out behavior, which relies on large magnitude turns designed to cross the plume. Fig. 7 clearly shows this behavior. Plume tracing for the mission shown in Fig. 8 primarily used the TrackIn behavior, since its small angle counterturning caused the AUV to drive up the main body of the plume. The difference in commanded altitudes could be the major reason for this difference, if the 2 m altitude of Fig. 7 only allowed the AUV to intermittently contact the top of the plume. Note also that in Fig. 8, as the AUV approaches the source, it must use the Track-Out behavior more frequently, because near the source the plume is still at a lower altitude.

The values of the parameters of the CPT strategy are summarized in Table 4. If P is increased, then the counterturns have a larger cross flow component. The tradeoff is that the larger crossflow component increases the probability that the AUV exits the plume from the expected edge (i.e., the variable LHS is more likely to be correct), but increases the length of the trajectory to get to the source. The variable X should be larger than the intermittent chemical detection gaps while "in the plume;" however, the plume intermittency is dependent on characteristics of the flow and turbulence that are not known. Typical "in the plume" interpulse durations are less than 1s (Jones, 1983). As X is increased, if chemical is not detected, then the distance that the AUV moves from the last detection point is increased. As long as this distance is less than Lu, then no backtracking is required. For these experiments, vc = 2.0 m/s. Therefore, for X = 5s, the distance traveled is 10 m which is less than Lu. The value of Lc was selected to ensure that, even with navigation errors (<10 m nominally) and with the Go To guidance command being satisfied when the AUV was within 10 m of the destination, the AUV would cross a line extending upflow from the last detection point. The value of N_re was set to 2. Increasing N_re causes the AUV to spend additional time searching upflow from each point on the lost_pnts list. This additional time is detrimental when the BowTie's are upflow from a false-alarm detection point. The values of K and R are dependent on the dynamic capabilities of the AUV. These values were determined in simulation and evaluated onboard the AUV prior to the CPT experiments described herein.

Note also, that the definition of a chemical detection implicitly contains two parameters: the detection threshold and the number of above threshold readings required to declare a detection. For all variations of CPT strategies that we performed during the three year program, the definition of a chemical detection was a concentration c(t)>4% of full scale (i.e., 0.2 V). This value was determined by analysis of chemical sensor data from the AUV operating in San Diego Bay (August 2002) in the absence of the chemical. In this scenario, the sensor readings were pure noise, but never surpassed 0.2 V. Therefore, we selected the threshold such that the probability of false alarm readings was extremely low. Therefore, any single sensor reading above threshold was registered as a chemical detection. The number of above threshold readings required to register a detection could be increased. This would decrease the probability of false alarms, but increase the probability of missed detections.

With the current AMP strategy and experimental results in mind, many alternative and possibly improved AMP strategies could be proposed. In fact, one of the goals of any experiment should be to identify areas for future improvements. Therefore, it is important to consider what lessons were learned in these experiments. First, care should be taken to ensure that the ADCP flow data corresponds to the flow layer containing the plume; however, this is not straightforward. For the Duck NC test location, the water is 4-8 m deep. The bottom boundary layer depth varied with time. The minimum safe AUV operating altitude was 1.5 m and the ADCP has an approximately 0.75 m deadzone prior to its measurement being accurate. Therefore, there were runs for which the upward looking ADCP was measuring the flow in the top layer instead of the bottom layer. Detecting and accommodating such events would require significant advancements for the planner and possible a conductivity, temperature, and depth (CTD) sensor with a fast response time. Second, some of the declared source locations had unexpected error in the crossflow direction, which was unexpected. We believe that this error component is due to navigation fixes that occurred near the time of declaration and by the declaration logic that ignored separation in the crossflow direction. The source declaration logic described herein was based only on the along flow separation of points at which the plume was lost. The crossflow separation was ignored in the declaration process to decrease the time required to make a declaration. Accounting for crossflow separation in the declaration logic would improve the accuracy of the declaration and is straightforward to implement in the future. Third, the current AMP strategy used the chemical sensor in a Boolean mode even though the sensor did provide an analog reading. It is often suggested that the analog concentration could provide a useful indicator of the distance to the source; however, there are a few difficulties in this approach. First, the chemical source concentration would be unknown in a real application. Second, the rate of decay of the peak concentration reading as a function of the distance from the source is flow dependent and not known. Third, maximum sensed concentration along any transect is not necessarily the maximum concentration in the vicinity of that transect. Alternative, the analog sensor reading could have utility in experiments where multiple sources might generate overlapping plumes. In that scenario, a significant decrease in the maximum sensed chemical while moving upflow might indicate that a source has just been passed by while the AUV is still in the plume of another source. Such strategies were not required for this project.

It is also interesting to consider adaptation of the AMP strategy parameters based on distance from the source. For example, it might be more efficient to decrease Lu and Lc as the AUV gets nearer to the source. The difficulty in implementing such ideas is in evaluating the distance to the source when the source location is unknown. Early in the program, we hoped that the width of plume transects would be a useful indicator of the distance to the source. This proved futile for a variety of reasons: plume meander results in AUV transects being at different angles relative to the plume centerline; a variety of factors result in AUV transects being at different altitudes relative to the plume centerline altitude; and, the instantaneous plume width at a fixed distance from the source varies widely. Similarly, sensed chemical concentration is not a useful indicator of distance to the source since the source concentration is unknown and the sensed concentration at a fixed distance from the source varies widely.

Learn Photoshop Now

Learn Photoshop Now

This first volume will guide you through the basics of Photoshop. Well start at the beginning and slowly be working our way through to the more advanced stuff but dont worry its all aimed at the total newbie.

Get My Free Ebook


Post a comment