Nội dung

Automotive Manufacturing: Intelligent Resistance Welding 229 Table 8. Fuzzy logic controller rule-base Rule 1 2 3 4 5 6 7 8 9 If If If If If If If If If “E” “E” “E” “E” “E” “E” “E” “E” “E” is is is is is is is is is low AND “N ” is low THEN dI = Pa medium AND “N ” is low THEN dI = Ng /2 high AND “N ” is low THEN dI = Ng low AND “N ” is medium THEN dI = Pa /2 medium AND “N ” is medium THEN dI = Ng /4 high AND “N ” is medium THEN dI = Ng /2 low AND “N ” is high THEN dI = Pa /4 medium AND “N ” is high THEN dI = Ng /8 high AND “N ” is high THEN dI = Ng /4 Fig. 7. Membership functions of the number of expulsion welds and the number of normal welds in a process window of the last p welds (p is a fixed parameter). Parameter p defines a universe [0, p] of the all possible expulsion and normal welds within that moving window the number of expulsion and normal welds in the last window of p welds through a set of rules with fuzzy predicates (Table 8). In the rules of the fuzzy logic controller low, medium, and high are fuzzy subsets defined on the [0, p] universe for the number of expulsions “E,” and the number of normal welds “N” (Fig. 7). Ng < 0 and Pa > 0 are constants (fuzzy singletons) defining control changes of the current. The first three fuzzy rules deal with the case where the number of normal welds “N” in the last window is low. Based on the number of detected expulsions, three alternative strategies for changing current level are considered: • • If the number of expulsions is low, it is reasonable to think that the state of the welds is close to the cold welds status. Hence, it is necessary to increase gradually the amount of current, i.e., the current is changed by. dI = Pa. If the number of detected expulsions is medium or high, it is reasonable to think that the state of the welds is close to the expulsion state. Hence, it is necessary to decrease the amount of current. This is performed selectively, based on the number of expulsions (high vs. medium), resulting in negative changes of the current dI = Ng vs. dI = Ng/2. When the number of normal welds N in the process window is medium, the strategies for adjusting the current level are as follows (rules 4–6): • • • When we have low expulsion detection rate, the weld state is likely approaching a cold weld. Therefore, the level of current should be increased. This is done by increasing the current level, i.e., dI = Pa/2. Note that the amount of increase when the number of normal welds “N” is medium (dI = Pa/2) is less than in the case when that number “N” is low (dI = Pa). The next case deals with medium expulsion rate, i.e., the weld state is close to the expulsion status. This requires a gradual reduction of the current dI. Note that the amount of decrease when “N” is medium (dI = Ng/4) is also less than the case when the “N” is low (dI = Ng/2). The last case appears when the expulsion rate is high. Since this is an undesirable state, the level of current should be lowered dramatically to minimize the number of expulsions. This is also done by modifying the secondary current dI = Ng/2 when “N” is medium and dI = Ng when “N” is low. 230 M. El-Banna et al. The last three fuzzy rules (7–9) consider high level of normal welds, i.e., satisfactory weld quality. Their corresponding control strategies are: • • • If we have low expulsion detection, the state of the welds will be approaching a cold weld status. Therefore, current level should be increased to prevent potential cold welds. This is done by a minor positive change of the current to dI = Pa /4. If we have medium expulsion detection, it is reasonable to consider that the state of the welds is close to the expulsion welds status. Therefore the current level should be decreased gradually to dI = Ng/8). In the last case, when the expulsion detection is high, the level of the current should be decreased. The corresponding change of the current is slightly negative (dI = Ng/4), i.e., significantly less than in the cases when “N” is medium (dI = Ng/2) or when “N” is low (dI = Ng). Applying the Simplified Fuzzy Reasoning algorithm [19], we obtain an analytical expression for the change of the current dI depending on the rates of expulsion welds “E” and normal welds “N” as follows:  µi (x)νj (y)∆i,j ∀i ∀j dI =   µi (x)νj (y) ∀i ∀j where: µi : membership function of the linguistic value of the expulsion welds {low, medium, high} νj : membership function of the linguistic value of the normal welds {low, medium, high} x: number of expulsion welds in the process window detected by the expulsion algorithm y: number of normal welds in the process window detected by the LVQ soft sensing algorithm µi (x): firing level for the expulsion membership function νj(y) : firing level for the normal membership function ∆i,j : amount of increment/decrement when the linguistic value of expulsion welds is “i” and the linguistic value of normal welds is “j” (for example, if the linguistic value of the expulsion welds is high and the linguistic value of the normal welds is low then ∆high,low = N g, where Ng negative value determines the change of the current dI); see Table 8. Triangular shape membership functions µi , νj are used in the fuzzy control algorithm (Fig. 7) to define the linguistic values of the numbers of expulsion and normal welds in the process window. These membership functions depend on the scalar parameters a, b, c as given by: ⎧ ⎫ 0, x≤a ⎪ ⎪ ⎪ ⎨ x−a a ≤ x ≤ b⎪ ⎬ µi , νj (x, y; a, b, c) = b−a c−x b ≤ x ≤ c⎪ ⎪ ⎪ ⎪ ⎩ c−b ⎭ 0, c≤x The new target current (Inew ) for the next window of p welds will be: Inew = Iold + dIIold where Iold is the current in the previous window of p welds and dI is the change of the current that is calculated from the fuzzy control algorithm. Proposed Intelligent Constant Current Control algorithm was implemented in Matlab/Simulink and was experimentally tested in a supervisory control mode in conjunction with an MFDC Constant Current Controller. Four sets of experiments were performed as follows. The first group of tests (with/without sealer) was performed using the Intelligent Constant Current Controller. The second group (with/without sealer) was carried out by using a conventional stepper. The role of the sealer in this test is to simulate a typical set of disturbances that are common for automotive weld processes. Sealer is commonly used to examine the performance of weld controllers and their capability to control process variability. Automotive Manufacturing: Intelligent Resistance Welding 231 Each group of tests consists of sixty coupons, i.e., 360 welds (for each test without sealer), and ten coupons, i.e., 60 welds (for each test with sealer) with two metal stacks for each coupon are used for each test. Both tests involved welding 2.00 mm gage hot tip galvanized HSLA steel with 0.85 mm gage electrogalvanized HSLA steel. Thirty six coupons (216 welds) without a sealer between sheet metals and ten coupons (60 welds) with a sealer for each group of tests were examined. Cold and expulsion welds were checked visually in each coupon. The length of the moving window in the Intelligent Constant Current Controller algorithm was p = 10, i.e., the soft sensing of expulsion and normal welds was performed on a sequence of ten consecutive welds. The negative and positive consequent singleton values in the rule-base of the fuzzy control algorithm were set at Ng = −0.09 and Pa = +0.07. In the stepper mode test, an increment of one ampere per weld was used as a stepper for this test. The initial input current was set at 11.2 kA for all tests, with no stabilization process to simulate the actual welding setup conditions in the plant after tip dressing. 4.1 Intelligent Constant Current Control and Stepper Based Control without Sealer Figure 8 shows the weld secondary current generated by the Intelligent Constant Current Control algorithm without sealer. It can be seen that at the beginning of the welding process, there were a couple of cold welds, so the fuzzy control scheme increased the current gradually until expulsion began to occur. When expulsion was identified by the soft sensing algorithm, the fuzzy control algorithm began to decrease the current level until expulsion was eliminated and normal welds were estimated again. After that it continued to increase the current until expulsion occurred again and so on. It can be concluded from the test above that the secondary current in the intelligent control scheme was responding to the weld status; in case of expulsion welds, the secondary current was decreased, and in case of cold welds, the secondary current was increased. Thus, the fuzzy control scheme was able to adapt the secondary current level to weld state estimated by the soft sensing algorithms. Figure 9 shows the secondary current in the case of conventional stepper mode. The weld primary current was set to a constant value at the beginning of the test, and then an increment of one ampere per weld was used as a stepper to compensate for the increase in electrodes diameter (mushrooming of the electrode); observed nonlinearity of the secondary current is a result of the transformer nonlinearities. It can be seen that there were several cold welds at the beginning of the test, followed by some of normal welds, and then expulsion welds were dominant until the end of the test. Evidently, the secondary current in the stepper mode was too aggressive towards the end of the welding process, resulting in many expulsion welds. On the other hand, at the beginning, the secondary current was not enough, resulting in cold welds. The stepper mode does not really adapt the current to the actual weld state at the beginning or at the end of the welding process. Fig. 8. Secondary current using the intelligent constant current fuzzy control algorithm 232 M. El-Banna et al. Fig. 9. Secondary current for the stepper based algorithm without sealer Table 9. Number of expulsion welds for the fuzzy control algorithm and the conventional stepper mode without sealer Number of expulsion welds using fuzzy controller Number of expulsion welds using stepper mode 68/216 = 31.5% 98/216 = 45.4% Table 10. Number of cold welds for the fuzzy control algorithm and the conventional stepper mode without sealer Number of cold welds using fuzzy controller Number of cold welds using stepper mode 31/216 = 14.4% 44/216 = 20.4% Tables 9 and 10 show the number of expulsion and cold welds for the Intelligent Constant Current Control algorithm versus the conventional stepper mode implementation. As expected, the number of expulsion welds in the stepper mode (98/216 = 45.4%) is higher than the number of expulsion welds in the fuzzy control scheme (68/216 = 31.5%). It can also be seen that the number of cold welds in the fuzzy control scheme test (31/216 = 14.4%) was less than the number of cold welds in the stepper mode (44/216 = 20.4%). 4.2 Intelligent Constant Current Control and Stepper Based Control with Sealer It is a common practice in the automotive industry to intentionally introduce sealer material between the two sheet metals to be welded. The purpose of this sealer is to prevent water from collecting between the sheets and in turn reduce any potential corrosion of the inner surface of sheet metals. However, the sealer creates problems for the spot welding process. In particular, the sealer increases the resistance significantly between the two sheet metals to be welded. When the welding process starts, high current will be fired, which is faced by high resistance (because of the sealer) in the desired spot to be welded, that prevents the current from flowing in that direction. The other alternative direction for this current is to flow in the direction of less resistance; this is what is known as shunting effect. Shunting effect produces cold welds, or at least small welds, which will cause a serious problem to the structure. Figure 10 shows the spot secondary current for the Intelligent Constant Current Control algorithm with sealer. It demonstrates a performance similar to the case with no sealer – increasing/decreasing of the current level to adapt to the estimated cold/expulsion welds. Figure 11 shows the spot stepper mode secondary current in the presence of sealer. The weld secondary current was set to a constant value at the beginning of the test with subsequent increments of one ampere per weld. It can be seen that the cold welds were dominant until just before the end of the test. There were a Automotive Manufacturing: Intelligent Resistance Welding 233 Fig. 10. Spot secondary current for the fuzzy control algorithm with sealer Fig. 11. Spot secondary current for the stepper mode with sealer Table 11. Number of expulsion welds for the fuzzy control scheme, the stepper, and the no stepper modes with sealer Number of expulsion welds using fuzzy controller Number of expulsion welds using stepper mode 3/60 = 5% 0/60 = 0.0% Table 12. Number of cold welds for the fuzzy control algorithm, the stepper, and the no stepper modes with sealer Number of cold welds using fuzzy controller Number of cold welds using stepper model 14/60 = 23.3% 43/60 = 71.7% couple of normal welds towards the end of the test. No expulsion welds occurred in this test. Apparently, the secondary current was not enough to produce cold welds. Using stepper mode does not adapt the secondary current according to the weld status. Tables 11 and 12 compare the number of expulsion and cold welds for the Intelligent Constant Current Control algorithm and the conventional stepper mode implementation in the case of welding with sealer. The number of expulsion welds in the fuzzy control scheme test (3/60 = 5%) is slightly higher than the number 234 M. El-Banna et al. of expulsion welds in the stepper mode test (0/60 = 0.0%). However, the number of cold welds in the case of application of the fuzzy control algorithm (14/60 = 23.3%) is much less than the number of cold welds in the conventional stepper mode test (43/60 = 71.7%). 5 Conclusions The problem of real time estimation of the weld quality from the process data is one of the major issues in the weld quality process improvement. This is particularly the case for resistance spot welding. Most of the models offered in the literature to predict nugget diameter from the process data employ measurements such as ultrasonics, displacement, and thermal force and are not suitable in an industrial environment for two major reasons: the input signals for prediction model are taken from intrusive sensors (which will affect the performance or capability of the welding cell), and, the methods often required very large training and testing datasets. In order to overcome these shortcomings, we proposed a Linear Vector Quantization (LVQ) neural network for nugget quality classification that employs the easily accessible dynamic resistance profile as input. Instead of estimating the actual weld nugget size the algorithm provides an on-line estimate of the weld quality by classifying the vectors of dynamic resistance profiles into three classes corresponding to normal, cold, and expulsion welds. We also demonstrated that the algorithm can be successfully applied when the dynamic resistance profile vector is replaced by a limited feature set. Based on the results from LVQ, a control algorithm called the Intelligent Constant Current Control for Resistance Spot Welding was proposed for adapting the weld current level to compensate for electrode degradation in resistance spot welding. The algorithm employs a fuzzy logic controller using a set of engineering rules with fuzzy predicates that dynamically adapt the secondary current to the state of the weld process. A soft sensor for indirect estimation of the weld quality employing an LVQ type classifier was implemented in conjunction with the intelligent control algorithm to provide a real time approximate assessment of the weld nugget status. Another soft sensing algorithm was applied to predict the impact of the current changes on the expulsion rate of the weld process. By maintaining the expulsion rate just below a minimal acceptable level, robust process control performance and satisfactory weld quality were achieved. The Intelligent Constant Current Control for Resistance Spot Welding was implemented and experimentally validated on a Medium Frequency Direct Current (MFDC) Constant Current Weld Controller. Results were verified by benchmarking the proposed algorithm against the conventional stepper mode constant current control. In the case when there was no sealer between sheet metal, it was found that the proposed intelligent control approach reduced the relative number of expulsion welds and the relative number of cold welds by 31% (from absolute 45.4–31.5%) and 29% (from absolute 20.4–14.4%) respectively, when compared to the stepper mode approach. In the case when there was a sealer type disturbance, the proposed control algorithm once again demonstrated robust performance by reducing the relative number of cold welds by 67% compared to the stepper mode algorithm (from absolute 71.7–23.3%), while increasing the absolute number of expulsion welds by only 5%. Our Intelligent Constant Current Control Algorithm is capable of successfully adapting the secondary current level according to weld state and to maintain a robust performance. Our focus in this chapter was on the Medium Frequency Direct Current (MFDC) weld controllers. An alternative version of the Intelligent Constant Current Control Algorithm that is applicable to the problem of alternating current (AC) weld control in conjunction with the Constant Heat Control Algorithm [8] is under development. References 1. M. Jou, Real time monitoring weld quality of resistance spot welding for the fabrication of sheet metal assemblies, Journal of Materials Processing Technology, 132, 102–113, 2003. 2. Y. Nishiwaki and Y. Endo, Resistance welding Controller, W. T. Corporation, Ed. USA, April 10, 2001. Automotive Manufacturing: Intelligent Resistance Welding 235 3. http://www.romanmfg.com/ 4. Y. Cho and S. Rhee, New Technology for measuring dynamic resistance and estimating strength in resistance spot welding, Measurement Science & Technology, 11, 1173–1178, 2000. 5. S. R. Lee, Y. J. Choo, T. Y. Lee, M. H. Kim, and S. K. Choi, A quality assurance technique for resistance spot welding using a neuro-fuzzy algorithm, Journal of Manufacturing Systems,20, pp. 320–328, 2001. 6. P. Podrzaj, I. Polajnar, J. Diaci, and Z. Kariz, Expulsion detection system for resistance spot welding based on a neural network, Measurement Science & Technology, 15, 592–598, 2004. 7. Y. Park and H. Cho, Quality evaluation by classification of electrode force patterns in the resistance spot welding process using neural networks, Proceedings of the Institution of Mechanical Engineers, Part B (Journal of Engineering Manufacture), 218, 1513, 2004. 8. M. El-Banna, D. Filev, and R. B. Chinnam, Online nugget quality classification using LVQ neural network for resistance spot welding, International Journal of Advanced Manufacturing Technology, 36, 237–248, 2008. 9. T. Kohonen, Improved versions of learning vector quantization. In: (2nd ed.), Proceedings of International Joint Conf. on Neural Networks I, pp. 545–550 San Diego, 1990. 10. N. T. Williams and J. D. Parker, Review of resistance spot welding of steel sheets Part 2 – Factors influencing electrode life, International Materials Reviews, 49, 77–108, 2004. 11. N. T. Williams, R. J. Holiday, and J. D. Parker, Current stepping programmes for maximizing electrode campaign life when spot welding coated steels, Science and Technology of Welding and Joining, 3, 286–294, 1998. 12. R. W. Messler, J. Min, and C. J. Li, An intelligent control system for resistance spot welding using a neural network and fuzzy logic, presented at IEEE Industry Applications Conference, Orlando, FL, USA, Oct. 8–12, 1995. 13. X. Chen and K. Araki, Fuzzy adaptive process control of resistance spot welding with a current reference model, presented at Proceedings of the IEEE International Conference on Intelligent Processing Systems, ICIPS, Beijing, China, 1998. 14. S. Lee, Y. Choo, T. Lee, M. Kim, and S. Choi, A quality assurance technique for resistance spot welding using a neuro-fuzzy algorithm, Journal of Manufacturing Systems, 20, 320–328, 2001. 15. M. El-Banna, D. Filev, and R. B. Chinnam, Intelligent Constant Current Control for Resistance Spot Welding, Proceedings of 2006 IEEE World Congress of Computational Intelligence, 2006 IEEE International Conference on Fuzzy Systems, Vancouver, 1570–1577, 2006. 16. D. Dickinson, J. Franklin, and A. Stanya, Characterization of Spot-Welding Behavior by Dynamic Electrical Parameter Monitoring, Welding Journal, 59, S170–S176, 1980. 17. M. Hao, K. A. Osman, D. R. Boomer, C. J. Newton, and P. G. Sheasby, On-line nugget expulsion detection for aluminum spot welding and weldbonding, SAE Transactions Journal of Materials and Manufacturing, 105, 209–218, 1996. 18. H. Hasegawa and M. Furukawa, Electric Resistance Welding System, U.S Patent 6130369, 2000. 19. R. R. Yager and D. Filev, Essentials of Fuzzy Modeling & Control. Wiley, New York, 1994. 20. T. Kohonen, Self-Organization and Associative Memory, 2nd ed., Springer, Berlin Heidelberg New York, 1987. Intelligent Control of Mobility Systems James Albus, Roger Bostelman∗ , Raj Madhavan, Harry Scott, Tony Barbera, Sandor Szabo, Tsai Hong, Tommy Chang, Will Shackleford, Michael Shneier, Stephen Balakirsky, Craig Schlenoff, Hui-Min Huang, and Fred Proctor Intelligent Systems Division, National Institute of Standards and Technology (NIST), 100 Bureau Drive, Mail Stop 8230, Gaithersburg, MD 20899-8230, USA 1 Introduction The National Institute of Standards and Technology (NIST) Intelligent Control of Mobility Systems (ICMS) Program provides architectures and interface standards, performance test methods and data, and infrastructure technology needed by the U.S. manufacturing industry and government agencies in developing and applying intelligent control technology to mobility systems to reduce cost, improve safety, and save lives. The ICMS Program is made up of several areas including: defense, transportation, and industry projects, among others. Each of these projects provides unique capabilities that foster technology transfer across mobility projects and to outside government, industry and academia for use on a variety of applications. A common theme among these projects is autonomy and the Four Dimensional (3D + time)/Real-time Control System (4D/RCS) standard control architecture for intelligent systems that has been applied to these projects. NIST’s Intelligent Systems Division (ISD) has been developing the 4D/RCS [1, 2] reference model architecture for over 30 years. 4D/RCS is the standard reference model architecture that ISD has applied to many intelligent systems [3–5]. 4D/RCS is the most recent version of RCS developed for the Army Research Lab (ARL) Experimental Unmanned Ground Vehicle program. ISD has been applying 4D/RCS to the ICMS Program for defense, industry and transportation applications. The 4D/RCS architecture is characterized by a generic control node at all the hierarchical control levels. Each node within the hierarchy functions as a goal-driven, model-based, closed-loop controller. Each node is capable of accepting and decomposing task commands with goals into actions that accomplish task goals despite unexpected conditions and dynamic perturbations in the world. At the heart of the control loop through each node is the world model, which provides the node with an internal model of the external world. The fundamental 4D/RCS control loop structure is shown in Fig. 1. The world model provides a site for data fusion, acts as a buffer between perception and behavior, and supports both sensory processing and behavior generation. In support of behavior generation, the world model provides knowledge of the environment with a range and resolution in space and time that is appropriate to task decomposition and control decisions that are the responsibility of that node. The nature of the world model distinguishes 4D/RCS from conventional artificial intelligence (AI) architectures. Most AI world models are purely symbolic. In 4D/RCS, the world model is a combination of instantaneous signal values from sensors, state variables, images, and maps that are linked to symbolic representations of entities, events, objects, classes, situations, and relationships in a composite of immediate experience, short-term memory, and long-term memory. Real-time performance is achieved by restricting the range and resolution of maps and data structures to what is required by the behavior generation module at each level. Short range, high resolution maps are implemented in the lower levels, with longer range, lower resolution maps at the higher levels. A world modeling process maintains the knowledge database and uses information stored in it to generate predictions for sensory processing and simulations for behavior generation. Predictions are compared with ∗ Corresponding author, roger.bostelman@nist.gov J. Albus et al.: Intelligent Control of Mobility Systems, Studies in Computational Intelligence (SCI) 132, 237–274 (2008) c Springer-Verlag Berlin Heidelberg 2008 www.springerlink.com
238 J. Albus et al. Mission (Goal) SENSORY PROCESSING Classification Estimation Computation Grouping Windowing WORLD MODELING VALUE JUDGMENT KNOWLEDGE BEHAVIOR GENERATION Maps Entities Task Knowledge Images Events Planners Executors internal external Sensors World Actuators Fig. 1. The fundamental structure of a 4D/RCS control loop observations and errors are used to generate updates for the knowledge database. Simulations of tentative plans are evaluated by value judgment to select the “best” plan for execution. Predictions can be matched with observations for recursive estimation and Kalman filtering. The world model also provides hypotheses for gestalt grouping and segmentation. Thus, each node in the 4D/RCS hierarchy is an intelligent system that accepts goals from above and generates commands for subordinates so as to achieve those goals. The centrality of the world model to each control loop is a principal distinguishing feature between 4D/RCS and behaviorist architectures. Behaviorist architectures rely solely on sensory feedback from the world. All behavior is a reaction to immediate sensory feedback. In contrast, the 4D/RCS world model integrates all available knowledge into an internal representation that is far richer and more complete than is available from immediate sensory feedback alone. This enables more sophisticated behavior than can be achieved from purely reactive systems. A high level diagram of the internal structure of the world model and value judgment system is also shown in Fig. 1. Within the knowledge database, iconic information (images and maps) is linked to each other and to symbolic information (entities and events). Situations and relationships between entities, events, images, and maps are represented by pointers. Pointers that link symbolic data structures to each other form syntactic, semantic, causal, and situational networks. Pointers that link symbolic data structures to regions in images and maps provide symbol grounding and enable the world model to project its understanding of reality onto the physical world. Figure 2 shows a 4D/RCS high level diagram duplicated many times, both horizontally and vertically into a hierarchical structure as applied to a single military vehicle (lowest level) through an entire battalion formation (highest level). This structure, now adopted as a reference model architecture for the US Army Future Combat System, among other organizations, could also be applied to civilian on-road single or multiple vehicles as information could be passed from one vehicle to the next or to highway communication and control infrastructure. This chapter will briefly describe recent project advances within the ICMS Program including: goals, background accomplishments, current capabilities, and technology transfer that has or is planned to occur. Several projects within the ICMS Program have developed the 4D/RCS into a modular architecture for intelligent mobility systems, including: an Army Research Laboratory (ARL) Project currently studying onroad autonomous vehicle control, a Defense Advanced Research Project Agency (DARPA) Learning Applied to Ground Robots (LAGR) Project studying learning within the 4D/RCS architecture with road following application, and an Intelligent Systems Ontology project that develops the description of intelligent vehicle behaviors. Within the standards and performance measurements area of the ICMS program, a Transportation Project is studying components of intelligent mobility systems that are finding their way into commercial crash warning systems (CWS). In addition, the ALFUS (Autonomy Levels For Unmanned Systems) project determines the needs for metrics and standard definitions for autonomy levels of unmanned systems. And a JAUS (Joint Architecture for Unmanned Systems) project is working to set a standard for interoperability between components of unmanned robotic vehicle systems. Testbeds and frameworks underway at NIST include the PRIDE (Prediction in Dynamic Environments) framework to provide probabilistic predictions of a moving object’s future position to an autonomous vehicle’s planning system, as well as the Battalion Formation SP Platoon Formation WM BG SURROGATE BATTALION Plans for next 24 hours SP WM BG SURROGATE PLATOON Plans for next 2 hours Section Formation SP WM BG SURROGATE SECTION Objects of attention SP WM BG VEHICLE Communication Attention Surfaces SP WM BG SP WM BG Plans for next 10 minutes Tasks relative to nearby objects Plans for next 50 seconds Task to be done on objects of attention Mission Package Locomotion SP WM SP WM BG BG SUBSYSTEM 5 second plans Subtask on object surface Obstacle-free paths Lines SP WM BG SP WM BG SP WM BG SP WM BG PRIMITIVE 239 OPERATOR INTERFACE Intelligent Control of Mobility Systems 0.5 second plans Steering, velocity Points SP WM BG SP WM BG SP WM BG SP WM BG SENSORS SP WM BG SP WM BG SP WM BG SP WM BG SERVO 0.05 second plans Actuator output AND ACTUATORS Fig. 2. The 4D/RCS Reference Model Architecture showing multiple levels of hierarchy USARSim/MOAST (Urban Search and Rescue Simulation/Mobility Open Architecture Simulation and Tools) framework that is being developed to provide a comprehensive set of open source tools for the development and evaluation of autonomous agent systems. A NIST Industrial Autonomous Vehicles (IAV) Project provides technology transfer from the defense and transportation projects directly to industry through collaborations with automated guided vehicles manufacturers by researching 4D/RCS control applications to automated guided vehicles inside facilities. These projects are each briefly described in this Chapter followed by Conclusions and continuing work. 2 Autonomous On-Road Driving 2.1 NIST HMMWV Testbed NIST is implementing the resulting overall 4D/RCS agent architecture on an ARL High-Mobility Multipurpose Wheeled Vehicle (HMMWV) testbed (see Fig. 3). Early work has focused on the lower agent control modules responsible for controlling the speed, steering, and real-time trajectory paths based on sensed road features such as curbs. This effort has resulted in sensor-based, on-road driving along dynamically-smooth paths on roadways and through intersection turns [6]. Future work includes the implementation of selected driving and tactical behaviors to further validate the knowledge set. NIST has put in place several infrastructural elements at its campus to support the intelligent vehicle systems development described above. An aerial survey was completed for the entire site, providing high resolution ground truth data. A GPS base station was installed that transmits differential GPS correction data across the site. Testbed vehicles, equipped with appropriate GPS hardware, can make use of these corrections to determine their location in real-time with an uncertainty a few centimeters. This data can be collected and compared to the ground truth data from the survey to make possible extensive vehicle systems performance measurements of, for example, mobility control and perception components. In addition, lane markings consistent with the DOT Manual on Uniform Traffic Control Devices, have been added to parts of the campus to support development of perception capabilities required for urban driving.