Distributed Computing in Autonomous Vehicles

QUANTUM ACCELERATED COMPUTER ARCHITECTURE

Presented By:1. Ranjith Reddy Gaddam 2. Uday Teja Vunnam 3. Uday Kiran Parvathagiri 4. Nikhila Jonnalagadda

Index

Abstract

Functional components

Centralized System Architecture

Distributed System Architecture

Development Process for Distributed System

Conclusion

References

ABSTRACT

An autonomous car is a self-driving vehicle having the ability to sense its surroundings and navigate without the assistance of a human. Complex autonomous driving algorithms are needed for self-driving vehicles, together with a wide variety of heterogeneous sensors, actuators, and computers. These algorithms include perception, localization, planning, and control. To control the heterogeneity of the system's components and the complexity of the driving algorithms, It suggests a development procedure and a system platform for the distributed system of an autonomous car by applying distributed system architecture to the autonomous driving system. The criteria for creating and developing the distributed system of an autonomous vehicle are provided by the development process.

+ info

FUNCTIONAL COMPONENTS

An autonomous car is a self-driving car that can drive by itself without human intervention. The autonomous vehicle is powered by five fundamental processes:

• Perception
• localization
• planning
• control
• system administration

The diagram depicts the conceptual explanation of each function.

+ info

+ info

In order to implement the functional components on the computing units of the autonomous car, there are two types of approaches:

1. Centralized architecture
2. Distributed architecture

CENTRALIZED SYSTEM ARCHITECTURE

• In the centralized system architecture, most functional components of the autonomous driving system are implemented into a single computing unit.

• As depicted in the diagram, the autonomous driving system's sensors and actuators are all linked to a single, centralized computer.

• The centralized system architecture simplifies system configuration because all the components, including the sensors and actuators, are connected to the central computing system.

• Without an additional network, it is possible to transfer information with the least transmission latency and information loss.

1.Accelerating quantum computer developments

Considering the quantum computer as a product requires standardization and integration of all of its building blocks, as well as a mature supply chain capable of providing high-quality components and ensuring supply security. The product development approach focuses on the product's functionality and performance requirements and builds the product with breaking technology. A product development roadmap can be outlined based on the expected requirements of future products.A fully functional quantum computer is expected to be available within a decade and used by the High-Performance Computing (HPC) market, where it will replace (part of) the supercomputers currently used for complex calculations and data management. In the short term, a partially functional quantum computer will be available and of interest to the R&D market, which requires such a product to accelerate the development of quantum technology.

+ info

+ info

Fig: Quantum Computing System

PRODUCT DEVELOPMENT

• Product development prioritizes performance and usability.
• When designing a product, the performance, and functionality required by users determine the design decisions that must be made.
• The price users are willing to pay for the product has an impact.
• Consideration of the quantum computer as a product necessitates interface standardization and integration of all its building blocks, as well as integration of the quantum computer itself in a more extensive ICT system architecture.
• A quantum computer comprises a series of components that must work together to exploit quantum-mechanical phenomena such as superposition and entanglement.
• These quantum effects are delicate and difficult to control.
• Realizing the desired performance and functionality is a complex engineering challenge. Because of this complexity, and in order to develop this product efficiently, a systems engineering approach is recommended.

CONT

The two approaches are:

1. The first approach is to determine the product requirements in terms of quantum computer performance and functionality.
2. The second approach is to determine the product specifications required to meet these requirements.

PRODUCT REQUIREMENTS AND SPECIFICATIONS

Requirements: 1. Provide solutions for commercially interesting problems. 2. Enable the development and execution of NISQ applications. 3. Enable the development of quantum processors. 4. Tune-up the performance of Quantum Devices. Specifications:Considering the development of a quantum computer as a product requires a mature supply chain that can provide high-quality components while also ensuring supply security. Supply chain management is critical when developing a product. The current emerging supply chain provides enabling technologies and supporting component solutions that are sufficiently mature to scale far beyond quantum supremacy-level systems. However, some innovation bottlenecks remain, such as the production of high-quality quantum devices and overall system integration.

Product Road map

• Based on product requirements and specifications, it is possible to outline a product roadmap.
• A product roadmap describes how a product is likely to evolve in time based on the expected development of the underlying technologies, as well as customer needs.
• It is expected that technology will improve over time, although it will be hard to predict when each milestone will be reached. Quantum technology development is still in its embryonic phase and sudden step-changes in improvement of technology are likely to occur, which makes predictions hard. However, the use of the products based on quantum technology is better understood.
• It is expected that a fully functional quantum computer will be used by the High Performance Computing (HPC) market to do complex calculations and data management.
• Before that market can be serviced, partly functional quantum computers will already be of interest to players in the R&D market, that have a need for such a product to speed-up their quantum technology developments.
• Combining the expected technology developments and expected use of the technology leads to a product roadmap.
• We consider this roadmap largely generic and independent of the underlying quantum technology, although at times we might refer to specifics of a quantum device-based system for clarification

CONCLUSION

This research paper outlined how to use this technology to develop products. The product development approach focuses on the product's functionality requirements and builds it with breaking technology. This will speed up the development of the quantum computer as a commercially viable product. To build quantum computers that outperform classical computers, a series of simple quantum computers with specific functionalities is required. A roadmap for a quantum computer product line was provided.

+ info

Fig: Distributed system architecture of autonomous cars.

DEVELOPMENT PROCESS FOR DISTRIBUTED SYSTEM

A development procedure and a standard software platform was proposed for creating a distributed autonomous driving system. The proposed development process consists the following four steps:

1. AUTOSAR
2. Software component design
3. Computing unit mapping and
4. Implementation of function.

AUTOSAR

In the automotive field, there is an open and standardized automotive software architecture named AUTOSAR.(AUTOSAR is based on a component-based software design model for developing vehicular software and provides the methodology for designing the automotive software. Objectives of AUTOSAR :

• Scalability of the software to different vehicle and platform variants.
• Transferability of the system functions throughout the network.
• Integration of functional modules from various suppliers.
• Maintainability throughout the whole product life cycle.
• Software updates and upgrades over the product’s lifetime.

+ info

Fig: AUTOSAR Software Architecture

SOFTWARE COMPONENT DESIGN

• The software component represents the encapsulated part of the functionality in the autonomous driving application.

• At the software component design step, the software components of the autonomous driving functions are designed, and the flow of information between the components is defined.

• The connection for sharing information between the software components, sensors, and actuators can be defined by using a virtual functional bus (VFB). It has two types of communication.

i. Client–Server communication: The client is a service requester, whereas the server is a service provider, which waits for incoming requests from any client.ii. Sender-Receiver communication: The sender does not know how many receivers are in the network and provides the information without any requests.

Fig: Software Component Design

+ info

COMPUTING UNIT MAPPING

• The designed software components should be assigned to each distributed computing unit.

• Assigning the software components includes the following steps:

i. First, a computing unit in charge of the sensor or the actuator should be installed adjacent to the device and contain specific software components to operate the device. ii. Second, all mapped software components in one computing unit should work at a pre scheduled time. iii. Third, some software components might have operating system (OS) dependence due to device interfaces or available libraries. iv. Fourth, the amount of information exchanged between different computing units should not exceed the network capability.

Fig: Computing Unit Mapping

+ info

IMPLEMENTATION OF FUNCTIONS

• The software components should be implemented into an assigned computing unit platform by considering the computing performance, OS, sensor and actuator interface, and network.

• Therefore, the implementation of the software components may depend on the characteristics of the computing unit platform.

• This dependence can increase the cost of the implementation and maintenance of the software.

• To minimize the dependence problem of the software components, a common software platform, which can apply to different computing units, should be developed.

PROS

• Fault management of the autonomous driving system can become more convenient by using the distributed system architecture.

• It can be extremely dangerous if the single computing system fails due to faulty hardware or software if the autonomous system uses just one computing unit to manage all of the operations of autonomous driving.

• The distributed autonomous driving system has multiple computing modules, so each one can monitor the state of the others and provide a fallback for any failed operations.

• All the components in the distributed system are connected to each other. So components can easily share data with other nodes.

• More components can easily be added to the distributed system i.e. it can be scaled as required.

CONS

• Depending on the mission objectives, the whole autonomous driving algorithm can be huge and complex; as a result, a single computation unit might not be able to do all of the computations.

• Some messages and data can be lost in the network while moving from one component to another as the system is complex.

• The database connected to the distributed systems is quite complicated and difficult to handle as compared to a single user system.

• Overloading may occur in the network if all the components of the distributed system try to send data at once.

CONCLUSION

• A distributed system architecture is used for the system platform because it has many benefits for developing an autonomous driving system, such as reduction of the computational complexity of the entire system, fault-tolerant characteristics, and modularity of the system.

• The development process provides the guidelines to design and integrate the distributed systems of an autonomous car.

• A layered-architecture based software platform, which originated from AUTOSAR, is applied to the distributed system in order to improve the reusability, scalability, transferability, and maintainability of the application software.

REFERENCES

• https://www.researchgate.net/publication/265645639_Development_of_Autonomous_Car-Part_I_Distributed_System_Architecture_and_Development_Process

• https://ieeexplore.ieee.org/abstract/document/8328277/

• https://ieeexplore.ieee.org/document/8673972

• https://ieeexplore.ieee.org/abstract/document/8500436/

THANK YOU !