Designers are adding sensors and intelligent processing to their final systems to compensate for performance deficiencies, and these designs are cheaper to manufacture and operate. As microprocessor and sensor prices continue to fall, both automated and semi-automated systems can incorporate more intelligence, making more optimized decisions based on a better understanding of their internal state and the external environment. Adding sensors to a design and intelligently correlating the data from all these sensors increases design time and complexity, but design teams are increasingly accepting this cost because the trade-off can result in a differentiated system that provides more functionality more effectively at a price comparable to its predecessors. The use of sensors in embedded designs is not new. What's different is the increasing integration of sensors and processors into designs, from high-end automated systems to mass-produced consumer products. As sensor and processor prices continue to fall, the barriers to replacing mechanical control structures are constantly changing. The task of partitioning processing and correlating multiple sensors within the same system goes far beyond mechanical replacement, to the point that suppliers often consider their solutions proprietary information, as the correct combination of sensors and processing algorithms can achieve lower material costs, higher energy efficiency, and better system performance. The DARPA Urban Challenge and the Autonomous Vehicle Challenge have strongly demonstrated that fully autonomous vehicles equipped with numerous sensors can detect, analyze, predict, interact, and move within their respective environments (Reference 1). The challenges show that autonomous vehicles can drive and navigate entirely on their own, relying solely on various onboard sensors and positioning systems, without remote control or human drivers (Reference 2). However, in these challenges, administrators load the autonomous vehicles with a list of target GPS (Global Positioning System) locations, instructing them on where to go. They do not decide where to go or the order in which to reach those points; they only know to reach the various points on the list from their current location. The DARPA Urban Challenge, held on November 3, 2007, required autonomous vehicles to drive alongside other human and autonomous vehicles in urban traffic conditions to demonstrate their ability to safely perform complex maneuvers such as lane changes, overtaking, parking, and crossing intersections. Six teams successfully completed the DARPA Urban Challenge preliminary rounds. The winning vehicle, from the Tartan racing team, utilizes seven lidar (light detection and ranging), radar (radio frequency detection and ranging), and vision sensors (reference 3) in addition to inertial GPS/IMU (inertial measurement unit) sensors. The sensor selection supports data fusion for algorithm planning and achieves data redundancy and correlation through sensor overlap. Robots are a class of evolving automated systems with numerous sensors. For example, Boston Dynamics offers various robots, such as the remotely controlled BigDog, which can traverse challenging terrain (including ice) and retreat when encountering difficulties, relying on its own sensors and onboard control system. BigDog's motion sensors include: coupling position, coupling force, ground contact, ground load, a laser gyroscope, and a stereo vision system. Other sensors primarily monitor the system's internal health, such as hydraulic pressure, oil temperature, engine temperature, RPM, and battery charging. Supplier iRobot also offers various robot models, including consumer-grade robotic vacuum cleaners like Roomba. Roomba uses multiple IR (infrared) sensors to detect its environment directly (or via its mechanical wings) (Reference 4). Semi-automatic systems represent an emerging area of ​​dense sensor design. At the high end are fly-by-wire aircraft and automobiles, while at the low end, significant growth is seen in consumer appliances such as washing machines. Semi-automatic systems receive high-level instructions from human operators but are responsible for managing the low-level operational details of the systems requiring monitoring. In a sufficiently broad interpretation, most embedded systems fall into this category, and their designers may draw knowledge from other sensor designs or data-rich designs (see the appendix "Supercomputing"). Complex remote control systems (such as BigDog) must respond automatically and instantly to their environment and conditions, partly because a single remote control interface is insufficient to provide adequate data bandwidth and feedback, preventing operators from using multiple adjustment commands to keep the system functioning correctly. When design teams choose to implement a semi-automatic subsystem, they should aim to make it faster and better than most operators can perform the same task manually. Semi-automatic fly-by-wire systems used in aircraft replace the physical controls between the pilot and the aircraft with electrical interfaces. The control system receives commands from the pilot and then, based on sensor readings, determines the optimal action of the actuators, making the most ideal movement at each control point. In this scenario, the smartest control system aims to allow the pilot to focus on high-level aircraft control, while the flight control system manages the lower-level control of each subsystem; this approach frees up valuable cognitive cycles for the pilot to focus on issues that the flight control system cannot compensate for. Automobiles are also increasingly adopting this high-low layered control system, positioned between the driver and the vehicle's control subsystems, making them safer and more efficient (Reference 5). Examples of automated subsystems in automobiles include anti-lock braking systems (ABS), electronic stability control (ESC), traction control, yaw control, and collision mitigation systems such as intelligent restraint systems and airbags. In different environments, the driver may not perceive these control systems. The automotive engine management system demonstrates a semi-automatic embedded subsystem with densely packed sensors. In addition to monitoring the driver's pedal input, the system tracks numerous other internal data points, such as temperature, pressure, and the chemical composition of air, fuel, and exhaust gases within the system. It further integrates with other sensors to measure ignition, knocking, and crankshaft position, thereby optimizing engine output, fuel efficiency, emissions performance, and driving experience, and even adapting to alternative fuels. False alarms, besides being a consideration of design and implementation costs, are limited by the semi-automatic embedded control system's ability to resolve ambiguous and ambiguous situations. For an "enhanced" system, its value is significantly reduced if it issues too many false alarms or requires excessive human intervention because it may damage the final system. The decline in the reliability of Mercedes' electronic control systems in the early 2000s is an example, demonstrating that a system's inability to distinguish between ambiguous or ambiguous situations negatively impacts its overall value. In such cases, correlating more sensor data allows semi-automatic embedded control systems to make safer, more complex decisions as they become increasingly adept at judging and avoiding reactions to false alarms. A high-end car's collision detection relies on the combined operation of numerous sensors, such as long-range and short-range radar, IR, video, inertial, and ultrasonic sensors, to detect and verify necessary actions in response to potential or impending collisions. Each of these sensors provides information about the surrounding environment, and the control system can correlate this data with other sensors, filling in the gaps in each type of sensor to avoid unintended consequences, such as airbags being triggered by a pebble hitting the bumper. In automobiles, an increasing number of warning subsystems provide auxiliary alerts, such as distance and blind spot detection, interacting directly with the driver to provide information or assistance. These warning and response systems also rely on the correlation of multiple sensor inputs to avoid false alarms or incorrect responses to certain situations. For example, a distance detection subsystem can correlate data from visual, inertial, wheel position, and steering column position sensors and then alert the driver to avoid false alarms. The use of more sensors and processing intelligence embedded in control systems has brought about cost and design complexity issues. As suppliers address these issues, designers are incorporating more complex automated control systems into lower-cost designs, including mid-range consumer applications. Priyabrata Sinha, Principal Applications Engineer at Microchip, points out that home appliances are gradually moving away from state machine boxes and incorporating more sensors and intelligence into their decision loops. For example, modern washing machines can use three microcontrollers to manage the system and user interface. It's worth noting that the amount of flash memory (not the size of the processor architecture) is the most significant difference between dual-sensor and six-sensor designs. The additional memory allows the system to hold more code for the sensors, enabling the program code to correlate multiple inputs for use by more complex control algorithms. Ritesh Tyagi, Senior Product/Segment Manager at Renesas Technology, says that a key part of a "fast" response system is how designers pair and connect sensors and processors. A mid-priced refrigerator might use up to eight microcontrollers and multiple corresponding local sensors to provide customized and optimized control for each compartment, such as the meat and vegetable compartment. These implementation types are challenging the balance between centralized and distributed processing to provide greater reliability, meet tight power budgets, and simplify user interaction with home appliances. Unfortunately, the methods of using multiple types of sensor data and correlating them in control algorithms involve many sensitive patent issues for various companies. However, some advanced examples may still inspire you to explore ways to collect sensor information and correlate it with other information within the system to create better designs, thereby achieving new value-added features more efficiently at a lower total cost. Tips: * Correlational sensor processing is a leading and increasingly mature technology. * Sensors are becoming smarter as designers incorporate more processing power. * Various types of applications (from high-end to low-end systems) are benefiting from the use of more sets of sensors, enabling them to handle more complex usage scenarios. References * “DARPA Urban Challenge,” www.darpa.mil/grandchallenge. * Cravotta, Robert, “Operating alone,” EDN, Dec 5, 2005, pg 49, www.edn.com/article/CA6288032. * “Tartan Racing: A Multi-Modal Approach to the DARPA Urban Challenge,” April 13, 2007, www.darpa.mil/grandchallenge/TechPapers/Tartan_Racing.pdf * Cravotta, Robert, “Rummage through a Roomba,” EDN, March 15, 2007, pg 32, www.edn.com/article/CA6421379. * Cravotta, Robert, “Making vehicles safer by making them smarter,” EDN, June 8, 2006, pg 49, www.edn.com/ article/CA6339246.