Scribd
Upload
3 views

Assignment 2 Parallel Distributed

The document outlines an assignment focused on optimizing real-time data processing in autonomous vehicle systems using parallel computing techniques. It identifies critical tasks such as obstacle detection and path planning that can be parallelized, and discusses suitable techniques like GPU acceleration and multi-core processing to enhance performance. Additionally, it emphasizes the importance of analytical modeling and performance evaluation to quantify improvements in decision-making speed and safety.

Uploaded by

numl-f23-20997
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
3 views

Assignment 2 Parallel Distributed

The document outlines an assignment focused on optimizing real-time data processing in autonomous vehicle systems using parallel computing techniques. It identifies critical tasks such as obstacle detection and path planning that can be parallelized, and discusses suitable techniques like GPU acceleration and multi-core processing to enhance performance. Additionally, it emphasizes the importance of analytical modeling and performance evaluation to quantify improvements in decision-making speed and safety.

Uploaded by

numl-f23-20997
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

National University of Modern Languages, Rawalpindi

Department of Computer Science


Fall 2024

Name : Shahid Hameed


Roll No : Rc-132
Assignment : Parallel & Distributed Computing

Assignment 2 [CLO-2]

Problem Scenario: Autonomous Vehicle Systems

Scenario: Autonomous vehicle control systems must process vast amounts of real-time sensor
data to make quick and accurate decisions. During high-speed driving or complex traffic
scenarios, the system experiences delays in processing sensor inputs, which could lead to unsafe
driving decisions.

Task: Develop an optimization plan for improving the real-time data processing of an
autonomous vehicle system using parallel computing techniques. The plan should focus on
analytical modeling and performance evaluation of parallel data processing systems,
particularly for sensor data processing, to ensure timely and accurate decision-making.

Deliverables:

1. Identification of Critical Processing Tasks:


o Identify critical tasks in an autonomous vehicle system that can be parallelized
(e.g., obstacle detection, path planning).
o Analyze how delays in these tasks impact vehicle performance and safety.
2. Selection of Parallel Computing Techniques:
o Research suitable parallel computing techniques to optimize real-time sensor data
processing (e.g., GPU acceleration, multi-core processing).
o Explain why these techniques would improve response times during complex
driving scenarios.
Soultion:
: 1. Identification of Critical Processing Tasks:
Critical Tasks in Autonomous Vehicle Systems:

* Obstacle Detection: The system must process data from various sensors (cameras, LIDAR,
radar) to identify and classify objects in the vehicle's path. Parallelizing this task can
significantly speed up detection and response times.
* Path Planning: Determining the best route or maneuver requires real-time processing of
environmental data. Path planning can be divided into smaller computations to run in parallel,
reducing overall decision time.
Sensor Fusion: Combining data from multiple sensors to form a coherent understanding of the
environment. This task can be split across different processors to handle different sensor streams
simultaneously.
* Object Tracking: Continuous monitoring of the positions of dynamic objects (e.., other
vehicles, pedestrians). Parallel processing of multiple tracked objects can reduce delays in
updating object
* Vehicle Control Systems: Low-level controls like steering, braking, and acceleration based on
realtime inputs can be handled in parallel with other tasks to ensure quick responses.
Impact of Delays:

* Delays in Obstacle Detection can result in missed or late identification of hazards, leading to
potential collisions.
* In Path Planning, delays can cause poor decisions in maneuvering, especially in fast-changing
environments like busy highways or intersections.
* Delayed Sensor Fusion can produce an incomplete or outdated picture of the environment,
affecting all other critical decisions.
* Object Tracking delays can lead to inaccurate predictions of the positions of moving
objects,slower responses from Vehicle Control Systems can result in unsafe driving behavior,
such as delayed braking.

2. Selection of Parallel Computing Techniques:


Suitable Techniques for Parallel Computing:
* GPU Acceleration:
* GPUs (Graphics Processing Units) are ideal for tasks like image processing, which is essential
for camera-based obstacle detection. GPUs can process thousands of data points simultaneously,
reducing detection times significantly.
* For example, obstacle detection tasks that rely on deep learning models for object recognition
benefit greatly from GPU acceleration.
* Multi-core Processing (CPU-based):
* Multi-core processors can divide sensor data processing tasks across several cores. For
instance, one core could handle radar data while another processes camera inputs, with
synchronization occurring to maintain accurate sensor fusion.
This is particularly useful for tasks like path planning, where multiple cores can evaluate
different possible paths simultaneously.
* Distributed Computing:

In cases where an autonomous vehicle is connected to an external cloud or edge computing


infrastructure, some computational tasks can be offloaded to nearby servers, reducing the load on
the vehicle's onboard processors.
This approach is particularly helpful for non-time-critical tasks like high-resolution mapping or
advanced object classification.
Why These Techniques Improve Response Times:

* GPU Acceleration allows for faster processing of complex sensor data, such as images from
cameras or point clouds from LIDAR, which are computationally intensive.
* Multi-core Processing enables concurrent execution of different parts of the system, such as
obstacle detection, path planning, and vehicle control, thereby reducing bottlenecks in decision-
making
* Distributed Computing can offload heavier tasks to remote servers, freeing up onboard
resources for time-sensitive tasks.
Analytical Modeling and Performance Evaluation:

* Modeling: Analytical models such as Amdahl's Law and Gustafson's Law can be used to
estimate the performance improvements when using parallel computing techniques. These laws
can help quantify how much acceleration can be expected by parallelizing tasks and utilizing
additional cores or GPUs.
* Evaluation: Simulations can be set up to evaluate the system's performance under different
scenarios (e.g., high-speed highways vs. crowded city streets) and measure improvements in
sensor data processing times, system latency, and overall safety metrics.

You might also like