# RSS Exam
###### tags: `RWTH Course` `RSS`
[ToC]
## Ex 1. View into the future
### 1. Robots can be classified according to the norm ISO 8373:2012.

### 2. Explain the main differences between industrial and service robots in terms of application fields.

* Industrial robot: applied in industrial automation applications.
* Service robot: performs useful tasks for humans or equipments excluding industrial automation appications.

### 3. The “Measurement Chain” according to DIN 1319-1:1995

## Ex 2. Control and feedback control systems
### 1. Robot system

### 2. Specify three technical main characteristics of an industrial robot as its is depicted in the figure above.

### 3. Cascade control
* position
* velocity
* current

### 4. Impact of factors on position error

### 5. Transformation chain
* 注意標法

## Ex 3. Electromagnetic sensos
### 1. Fundamentals of EM
* 左邊differential 右邊integral

* Functionality: Variable resistors that use the concept of an adjustable voltage divider. Works accoding to Ohm's law. Resistance proportional to rotation angle / absolute distance.
* Purpose: it's a positioning sensor, measuring angle or distance.


---
### 2. Odometry
* Absolute:
1. maintains position information when power is removed, and the info is available immediately on applying power
2. System does not need to return to a calibration point

* Incremental:
1. does not report or keep track of absolute position
2. May need to move to a fixed reference point to initialize the position.

---


---

### 3. Optical odometry

## Ex 4. Capacitive and piezoelectric sensors in robotics
### 1. Capacitive sensor - Capacitance


---


### 2. Capacitive sensor - plate capacitor

### 3. Piezoelectric sensor - Accelerometer

## Ex 5. (L6) Ultrasonic & thermoelectric
### 1. Thermal expansion



### 2. Metallic resistance thermometer (Pt)


---

---
* Relatively good linearity (change of resistance almost linear with temperature change)
* Large measuring range

### 3. Thermocouple - Seeback

### 4. Ultrasonic

## Ex 6. (L7) Machine vision in robotics
### 1. Spatial filtering

New value = 188+178+201+197+168

### 2. Hough
* Classic hough transform can detect edges or lines. It can also detect other structures (ex: circles) if their parametric equation is known.
* Principle: transform edges pixels in image space into parameters space

---

---

---
### 3. Canny

## Ex 7. (L9) Data acquisition
### 1. Position of sensor in data acquisition system

### 4. Sampling - Nyquist
* The Nyquist frequency corresponds to half of the sampling frequency and must be greater than the highest frequency occurring in the signal, in order to ensure artifact-free sampling.

---

### 5. Sampling - Nyquist

---

### 6. Quantization error


---

### 7. DAC - binray weighted resistor


---

### 8. Sensor fusion - Motivation

* Sensor deprivation: breakdown of a senso element --> loss of perception on the desired object
* Limited spatial coverage: usually an individual sensor only covers a restricted region
* Limited temporal coverage: some sensors need a particular set-up time to perform & transmit measurement --> limiting the maxiimum freq of measurements
* Imprecision: the precision of a single sensor is limited
* Uncertainty: due to the sensor's limited view of the object.

---

### 9. Sensor fusion - Types

## Ex 8. (L8) Data preprocessing - filtering & noise removal
### 1. Signal Filtering


---

---

---

---
* pros:
1. better frequency response (narrower transition band & better attenuation, thus better approx. to ideal filter)
2. greater design flexibility

* cons:
1. Larger phase shift
2. More complex design
3. Higher costs (and longer development time)

### 2. Noise and signal extraction

* White noise:
:arrow_right: has a constant power spectral density (i.e., constant power spectrum). (i.e., equal amount of energy per frequency band).
:arrow_right: Amplitude does not necessarily distributed according to a Gaussian distribution (ex: maybe uniform, or some other non-Gaussian) Could be generated using a random number generator.
* Gaussian noise:
:arrow_right: Does not always have a constant power spectrum.
:arrow_right: the probability distribution of noise amplitude follows a Gaussian distribution.
* White Gaussian noise
has a flat power spectrum and also a Gaussian probability distribution.

---

---
For specific types of noises:
In general:
* Apply filters (ex: bandpass filtering)
* lock-in amplification could improve SNR. (but require prior knowledge of the signal and its characteristics)
---

:::danger


:::

## Ex 9. (L10) Signal transmission

### 1. Modulation - Motivation
* to adapt physical properties of msg signal optimally to the transmission channel, but without implications for the transmitted msg.

### 2. AM

---


---

* Synchronous modulation requires transmitter and receiver to be synchronous in phase all the time. However, it is challenging (ex: carrier signal may phase shift over time). If mismatch occurs, then the demodulated signal would be distorted.
* Solution: use asynchronous demodulation, which avoids this necessity. It simply uses envelope detector that connects the peaks of the signal and is a good approximation of the original signal.

---