Aula 5 - Sensores

Introduction to Sensors in Robotics

Overview of Sensors and Actuators

• The discussion begins with the importance of sensors and actuators as key components of robots, enabling them to perceive their environment.

• Emphasizes that a robot's ability to perceive (or sense) its surroundings is crucial for it to be considered a true robot.

Types of Sensors

• Introduces the concept of sensors as devices that extract information about both the system (robot) and its surrounding environment.

• Clarifies that sensor data can provide insights into internal parameters of the robot, not just external conditions.

Internal vs. External Sensors

• Defines "proprioceptive sensors" as those measuring internal states, such as battery charge levels, which are vital for autonomous operation.

• Discusses practical applications like drones returning to base when battery levels are low, highlighting the need for preserving robotic systems.

Examples of Proprioceptive Sensors

• Mentions various proprioceptive sensors including battery level indicators and wheel speed encoders used for maintaining consistent speeds during operation.

• Explains how wheel encoders work using optical sensors to measure rotation speed, aiding in smooth acceleration control.

Importance of Environmental Awareness

• Highlights additional proprioceptive sensors like temperature monitors for processors, emphasizing their role in preventing system failures.

• Introduces inertial sensors (gyroscopes), which help determine a robot's orientation in space—critical for stability and navigation.

External Sensors: Understanding the Environment

Types of External Information Gathered by Sensors

• Discusses external sensors that gather environmental data such as light levels or object distances to prevent collisions.

• Stresses the significance of absolute positioning through GPS or compasses, allowing robots to understand their location within a larger context.

Sensor Integration Challenges

• Notes modern trends where many sensors come integrated with microcontrollers that process data before sending it out.

Understanding Image Representation in Robotics

The Role of Cameras in Robots

• Cameras capture images, which are represented digitally. Each image consists of pixels that convey luminosity.

• The resolution of displays (e.g., Full HD 1920x1080) indicates the number of horizontal and vertical points, forming a matrix.

• A Full HD display has approximately 2 million pixels (1920 x 1080), essential for understanding image quality.

Measuring Luminosity

• Pixels represent luminosity, which can be measured in binary terms: black (0) and white (1).

• In grayscale images, each pixel's intensity contributes to varying shades of gray, with 256 possible values in an 8-bit representation.

Color Representation

• Color images use three primary colors: red, green, and blue (RGB). Each color channel typically uses 8 bits for intensity.

• This results in over 16 million possible colors (2^8 x 2^8 x 2^8), allowing for rich color representation.

Challenges in Object Detection

• When using a high-resolution camera on a robot, detecting specific objects like a blue chair involves processing complex data from millions of pixels.

• Reducing resolution simplifies processing; for instance, working with monochromatic images requires less computational power than full RGB.

Image Processing Techniques

• Applying filters helps identify obstacles without needing detailed color information. For example, edge detection can highlight object boundaries.

Robotics Sensor Overview

Converting Sensor Data into Useful Information

• The discussion begins with the need to convert sensor data into actionable information for robots, emphasizing the importance of effective data processing.

• Examples of sensors include odometers for wheel rotation speed and collision detection sensors, which can be simple buttons that trigger upon impact.

Types of Sensors in Robotics

• Collision sensors function as switches; when an object hits a bumper, it closes a circuit, indicating a collision.

• Distance sensors are crucial for detecting objects and can utilize sound (sonar) or light (LIDAR), each serving different applications in robotics.

Additional Sensor Types and Their Functions

• Various sensors like LDR (Light Dependent Resistor), microphones, and load cells measure environmental factors such as light intensity, sound levels, and weight respectively.

• Load cells change resistance based on applied pressure, allowing measurement of weight through electrical signal changes.

Motion and Orientation Sensors

• For measuring rotation in robotic arms, encoders or potentiometers can be used to track movement accurately.

• Inertial sensors typically combine accelerometers and gyroscopes to provide comprehensive motion tracking capabilities.

Environmental Detection Sensors

• Chemical sensors detect gases like CO2 or gas leaks; temperature sensors measure heat using thermometers or infrared technology.

• Pressure gauges (manometers) and altimeters are also discussed as essential tools for measuring atmospheric conditions.

Understanding Transducers in Sensors

• Most sensors act as transducers by converting environmental energy into electrical signals that can be processed digitally.

• For example, an LDR varies its resistance based on light exposure, affecting current flow which is then measured to determine light intensity.

Complexity of Data Processing in Robotics

• The simplicity of binary sensor outputs (like collision detection buttons being either pressed or not pressed) contrasts sharply with complex data from cameras that require significant computational resources.

Understanding Sensor Data Processing

The Challenge of Image Processing

• Processing data from a camera using a microcontroller is complex due to the large volume of data, even if the image size is small.

• Simpler sensors with less information can trigger actions in the physical world, such as reversing a robot when an interrupt switch is activated.

Sensor Information and World Representation

• Sensors like cameras provide data that represent the world at a specific moment rather than how it was previously.

• This leads to the "signal to symbol" problem, where raw sensor data must be interpreted into meaningful information (e.g., identifying colors).

Pre-processing Signals

• Some sensors offer pre-processing capabilities; for example, microphones can filter specific frequency ranges before outputting signals.

• Digital Signal Processing (DSP) is crucial in robotics for handling signal pre-processing and image processing tasks.

Types of Sensors: Active vs. Passive

• Active sensors emit energy to measure environmental conditions (e.g., sonar), while passive sensors detect existing energy without emitting anything (e.g., thermometers).

• Examples include ultrasonic distance sensors that emit sound waves and LiDAR systems that use light.

Characteristics of Sensors

• When selecting sensors, important characteristics include operational range and sensitivity to specific frequencies or wavelengths.

Understanding Sensor Fusion in Robotics

Introduction to Sensor Limitations

• The importance of working within error margins and sensory limitations is emphasized, highlighting the need for multiple sensors to compensate for inaccuracies.

Types of Sensor Fusion

Redundant Fusion

• Redundant fusion involves using identical sensors to improve information quality; for example, two sonar sensors can be used to detect collisions more accurately.

• Physical redundancy occurs when two identical sensors are placed close together, allowing the system to use the minimum distance reading for collision detection.

Logical Redundancy

• Logical redundancy uses different types of sensors that provide the same type of information; e.g., a sonar and a LiDAR sensor both measuring distances to potential obstacles.

Complementary Fusion

• Complementary fusion combines different types of data from various sensors aimed at achieving the same goal; for instance, using temperature readings alongside visual data to identify human presence versus animals.

Coordinated Fusion

• In coordinated fusion, the order of sensor activation matters. For example, detecting a person by voice first before activating a camera saves energy while ensuring accurate identification.

Attributes of Sensors

Field of View and Range

• The field of view refers to the area covered by a sensor. Different sensors have varying degrees (e.g., sonar vs. cameras), affecting their detection capabilities.

Accuracy and Precision

• Accuracy measures how correct a sensor's readings are (e.g., 90% accuracy means it correctly identifies obstacles most times). However, high accuracy alone isn't sufficient if a faulty sensor consistently reports false positives.

Resolution

Understanding Sonar and Sensor Limitations

Sonar Resolution and Detection Limits

• A sonar with a resolution of 1 cm can detect objects at specific distances (e.g., 9 cm, 10 cm) but cannot differentiate between closely spaced measurements (e.g., between 9.5 cm and 10.2 cm).

• The effectiveness of analog sensors depends on both the precision of the signal they deliver and the quality of the analog-to-digital converter in use.

Environmental Considerations for Sensors

• The application environment is crucial; sonar may struggle in areas with many walls due to sound bouncing, while laser-based sensors can fail if aimed at reflective surfaces like mirrors.

• Energy consumption is a significant factor in mobile robotics, especially when using active sensors that emit energy to measure distances.

Energy Management Strategies

• Continuous reading from sensors or frequent communication can drain battery life; thus, readings should be taken periodically or only when necessary to conserve energy.

• Low battery levels can lead to unreliable sensor readings due to insufficient power supply affecting measurement accuracy.

Bandwidth and Data Retrieval Rates

• Sensor bandwidth refers to how quickly data can be returned; this affects obstacle detection capabilities significantly.

• For example, a sensor operating at 4 Hz provides updates every 0.25 seconds, which could be too slow for timely obstacle detection if the robot moves faster than it can process data.

Common Sensors Used in Robotics

Basic Switches and Their Applications

• The simplest sensor is an interruptor switch that opens or closes contacts based on current flow; it’s used for applications like end-of-travel limits in robotic arms.

Advanced Measurement Techniques

• Encoders can track wheel rotations by using switches that count each complete turn, providing feedback on speed and position.

Light-Based Sensors

Light Sensors and Their Applications

Understanding Light Sensors

• A fixed resistor is used to measure light intensity, allowing for variations in resistance based on luminosity levels.

• The need for an Analog-to-Digital Converter (ADC) arises to convert analog signals from light sensors into digital numbers that processors can interpret.

• Active sensors like photodiodes detect specific wavelengths of light, such as infrared, functioning oppositely to LEDs.

Types of Light Sensors

• Two main types of active sensors are discussed: reflective sensors that emit and detect reflected light, and interruption sensors that detect objects by breaking a beam of emitted light.

• Infrared detection is preferred because humans cannot see it, minimizing interference from ambient light sources.

Challenges with Ambient Light

• External factors like sunlight can disrupt sensor functionality; for example, the Kinect device struggles in bright sunlight due to its reliance on infrared signals.

• Pulsing the infrared signal helps distinguish between different light sources and reduces false readings caused by ambient conditions.

Practical Applications

• Stroboscopic or modulated infrared lights are used in applications like elevator door presence detectors to avoid misreading sunlight as an object interrupting the beam.

Encoder Sensors Explained

• Encoders (or optical odometers), primarily used for measuring wheel position and speed, operate using emitted light. They help estimate robot positioning based on movement speed.

Limitations of Encoder Technology

• While encoders provide velocity data useful for estimating position, they can lead to significant errors if wheels slip or do not move as expected.

• This method of estimation is known as "Dead Reckoning," which may result in inaccurate location tracking over time.

Mechanism of Operation

• An encoder typically consists of an emitter (often infrared LED), a disk with holes, and a detector. Each hole passing through generates pulses indicating movement.

Understanding Angular Positioning and Sensor Technologies

Angular Position Detection Techniques

• The discussion begins with the concept of angular positioning, emphasizing the need to know the initial state when using a robot. A volatile position can lead to loss of value if not properly managed.

• The speaker introduces a method for detecting both speed and direction of rotation using two channels positioned at 90 degrees to each other. This allows for identifying four states based on which channel is activated first.

• A potentiometer is mentioned as another means of determining angular position. It consists of a resistive track that changes resistance based on the cursor's position, similar to volume controls in audio devices.

• By measuring resistance through a voltage divider circuit, one can determine how far along the resistive track the potentiometer has moved, thus indicating its position.

• More complex sensors exist beyond basic potentiometers; these include sonar technology used for distance measurement by analyzing sound wave propagation.

Sonar Technology Overview

• Sonar operates by emitting sound waves and measuring their time of flight—how long it takes for them to return after hitting an object. This helps estimate distances effectively.

• Human hearing limitations are discussed; sonar emits frequencies between 40 kHz and 180 kHz, which are inaudible to humans but detectable by animals like dogs.

• The effectiveness of sonar diminishes with distance due to signal decay; as sound travels further, it becomes weaker and harder to detect accurately.

• Two types of sonar sensors are described: simpler models with higher blanking times (period during which they ignore incoming signals), making them less effective at close range detection.

• Advanced sonar systems have lower blanking times and can detect objects closer than simpler models. They emit sounds intermittently rather than continuously to avoid interference in readings.

Limitations and Challenges in Sonar Detection

• The operation mechanism involves emitting sounds in bursts followed by waiting periods for echoes, which may lead to missed detections if obstacles are too close together.

• The "monsto" sensor is highlighted as an example that combines emission and reception capabilities into one unit, improving detection efficiency within a limited range (up to 5 meters).

• An explanation follows regarding acoustic axes—the path sound travels—and areas where detection may fail due to reflections or obstructions behind detected objects.

• Cross-talk issues arise when multiple sonars operate simultaneously in confined spaces, leading one sensor's emissions to be misinterpreted by another sensor as valid data from different sources.

Understanding Sensor Limitations and Technologies

Challenges with Sonar Sensors

• The proximity of walls can cause reflection issues, necessitating a 90-degree firing angle for sonar sensors to avoid simultaneous triggering.

• The effective distance of the sensor is impacted by the coverage angle; at 30 degrees, the perceived distance may be reduced due to angular distortion known as "foreshortening."

• To improve accuracy, it is recommended to take multiple readings (at least three) from the same sonar sensor to calculate an average position of detected obstacles.

• Battery charge affects sound emission strength; lower battery levels result in weaker sound waves that travel shorter distances, limiting obstacle detection capabilities.

• Sonar operates at low frequencies (50 Hz), which limits detection speed and requires sequential readings at specific angles.

Advancements with Lidar Technology

• Lidar uses laser technology instead of sound, allowing for faster data collection due to light's higher propagation speed compared to sound.

• A rotating mirror in lidar systems enables rapid emission and reception of laser beams across various angles, enhancing spatial coverage.

• Each lidar reading covers a wider angle (180° to 240°), but takes longer (50–100 ms), involving mechanical movement for accurate scanning.

• Detection effectiveness depends on surface characteristics; dark or distant objects may not reflect laser signals well, while smooth surfaces like glass can hinder detection entirely.

Understanding Camera Technologies

• Cameras typically use two types of sensors: CCD and CMOS. CMOS is more common due to its cost-effectiveness and simpler design using photodiodes for light intensity detection.

• While CMOS cameras are cheaper and suitable for portable devices, they tend to pick up more noise because their transistors are integrated directly into the pixel matrix.

• CCD cameras provide higher quality images with less noise but are generally used in professional settings such as medical imaging or satellite photography due to their complexity and cost.

Understanding Color Representation and Camera Technologies

The Basics of Color Representation

• The discussion begins with the concept of color saturation, which indicates how far a color is from its pure form. It also introduces the value (intensity) of the color.

• The color model discussed includes components: Red, Green, Blue (RGB), luminance (Y), and chroma (UV). This model helps in understanding how colors are represented digitally.

Luminance and Monochromatic Images

• Luminance is crucial for creating monochromatic images; it allows for conversion without needing to process full-color data when working with black-and-white images.

• For applications requiring color information, the U and V components can be utilized to extract blue and red projections respectively.

Specialized Cameras and Their Applications

• The conversation shifts to specialized cameras like those that capture near-infrared light. These cameras are particularly useful in agriculture and mobile devices for features like facial recognition.

• Near-infrared technology detects heat emitted by living beings, allowing these cameras to differentiate between a person and a photograph effectively.

Stereoscopic Cameras Explained

• Stereoscopic cameras feature dual lens systems that capture two slightly offset images, providing depth perception.