About this previous year paper
Subject Code: BEC053
Maximum Marks: 70
Time: 3 Hours
Paper ID: 310954
Question Paper Overview
SECTION A (2 × 7 = 14 Marks)
(Short conceptual questions — fundamental robotics and AI concepts)
a. Differentiate between mobile ground robots and uninhabited air vehicles (UAVs).
b. Define stability in the context of control systems.
c. Define fuzzy logic in robotics.
d. Define Bayesian belief networks (BBNs) and their applications.
e. Differentiate between genetic algorithms (GA) and simulated annealing (SA).
f. Define a feed-forward neural network and state its applications.
g. Explain the concept of a decision tree.
SECTION B (Attempt any three × 7 = 21 Marks)
a. Compare and contrast mobile ground robots with uninhabited air vehicles, focusing on design and control challenges.
b. Explain probabilistic path planning, detailing its process and advantages in robot navigation.
c. Explain the role of probability and error models in measurement systems, with examples.
d. Explain decision tree structure and its use in classification problems, with examples.
e. Compare genetic algorithms, simulated annealing, and particle swarm optimization (PSO) in terms of methodology and application.
SECTION C (Attempt one part from each question × 7 = 35 Marks)
Q3
(a) Discuss biological and cognitive paradigms used in robot design.
OR
(b) Explain the Denavit–Hartenberg (D–H) transformation with an example for a two-joint robotic arm.
Q4
(a) Explain the principles of open-loop and closed-loop control systems with examples from robotics.
OR
(b) Explain the significance of stability and performance analysis in control systems.
Q5
(a) Explain the Extended Kalman Filter (EKF) and Particle Filter (PF) techniques with applications in robotics.
OR
(b) Explain the role of sensors in robotics. Describe the types of sensors and their integration in sensor-based estimation.
Q6
(a) Discuss Bayesian belief networks and their application in robot decision-making.
OR
(b) Explain the process of task planning for individual and multiple agents in robotics, with examples.
Q7
(a) Explain forward chaining and backward chaining with examples of their use in robotics.
OR
(b) Discuss the implementation of deep learning algorithms in robotics, with specific examples of applications.
Key Topics for Revision
1. Mobile Robots vs UAVs
| Aspect | Mobile Ground Robots | Uninhabited Air Vehicles (UAVs) |
|---|---|---|
| Environment | Land | Air |
| Mobility | Wheeled, legged, or tracked | Flying (rotor or fixed wing) |
| Control Challenges | Surface friction, terrain | Wind, altitude, stability |
| Sensors | LiDAR, GPS, Ultrasonic | IMU, GPS, Barometer, Camera |
2. Control System Stability
Definition: The ability of a system to return to equilibrium after disturbance.
Stable system: Output remains bounded for bounded input.
Tools: Root locus, Bode plots, Nyquist criterion.
3. Fuzzy Logic in Robotics
Mimics human reasoning using degrees of truth (0–1).
Used for navigation, obstacle avoidance, control decisions.
Example: A fuzzy controller adjusting motor speed based on proximity sensors.
4. Bayesian Belief Networks (BBNs)
Probabilistic graphical model showing dependencies among variables.
Used in robot perception, fault diagnosis, decision-making.
Example: Autonomous car assessing road safety using sensor uncertainty.
5. Genetic Algorithms (GA) vs Simulated Annealing (SA)
| Feature | GA | SA |
|---|---|---|
| Inspiration | Natural selection | Metallurgical annealing |
| Operators | Selection, crossover, mutation | Random neighbor exploration |
| Convergence | Parallel population search | Single-solution stochastic optimization |
| Applications | Path planning, design | Optimization, scheduling |
6. Feed-Forward Neural Networks (FFNN)
Data flows one way: input → hidden → output.
Used in pattern recognition, object detection, sensor fusion.
Example: Classifying object types from vision input.
7. Decision Trees
Tree-based model used for classification or regression.
Splits dataset based on information gain or Gini index.
Example: Robot deciding between actions: move, turn, stop.
8. Probabilistic Path Planning
Uses random sampling and probability models (e.g., PRM, RRT).
Efficient for high-dimensional robot navigation.
Advantages: Handles uncertainty, adaptable to dynamic environments.
9. Kalman Filters
| Type | Purpose | Use in Robotics |
|---|---|---|
| EKF (Extended) | Non-linear motion model | Localization, SLAM |
| Particle Filter | Monte Carlo sampling approach | Mobile robot tracking |
10. Sensors in Robotics
| Type | Example | Function |
|---|---|---|
| Proximity | Ultrasonic, IR | Detect obstacles |
| Vision | Camera, LiDAR | Mapping, recognition |
| Inertial | Gyroscope, accelerometer | Orientation, balance |
| Force/Torque | Strain gauge | Interaction feedback |
11. Denavit–Hartenberg (D–H) Transformation
Standardized method to represent joint parameters (θ, d, a, α).
Used in robot arm kinematics.
Example: For a 2-joint arm, compute transformation matrix
- T=A1×A2T = A_1 \times A_2T=A1×A2
where each AiA_iAi is a 4×4 homogeneous transformation.
12. Biological & Cognitive Paradigms
Biological: Inspired by animal locomotion (e.g., insect robots, swarm robotics).
Cognitive: Inspired by human reasoning (AI-based planning, learning).
13. Control Systems in Robotics
| Type | Description | Example |
|---|---|---|
| Open Loop | No feedback | Simple motor control |
| Closed Loop | Feedback-based correction | Line-following robot |
| Performance Analysis: Includes transient response, steady-state error, and stability margin. |
14. Task Planning
Individual Agent: Optimize one robot’s path or task (warehouse robot).
Multi-Agent: Coordination and scheduling between multiple robots.
Techniques: Graph-based search, AI planning (STRIPS), reinforcement learning.
15. Forward & Backward Chaining
| Type | Approach | Use |
|---|---|---|
| Forward Chaining | From facts → conclusions | Rule-based control |
| Backward Chaining | From goal → facts | Diagnostic reasoning |
16. Deep Learning in Robotics
CNNs: Object detection, vision-based control.
RNNs/LSTMs: Sequence prediction, motion control.
Reinforcement Learning: Autonomous decision-making (e.g., Deep Q-learning in navigation).
Applications:
Self-driving cars Industrial robot arms
Human-robot interaction Drone navigation
Exam Preparation Tips
Draw D–H parameter tables and transformation matrices clearly.
Review EKF/Particle Filter equations for state estimation.
Include block diagrams for fuzzy logic, control loops, and BBNs.
Use real-world robotics examples (e.g., drones, manipulators, autonomous cars).
Prepare short, precise definitions for Section A, and descriptive answers with flowcharts/diagrams for Sections B and C.