Remote Assistance or Remote Driving: The Impact of Operational Design Domains on ADS-Supporting Systems Selection

Ole Hans, Benedikt Walter · 2025

Read Remote Assistance or Remote Driving: The Impact of Operational Design Domains on ADS-Supporting Systems Selection by Ole Hans, Benedikt Walter (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-14347.txt.

Introduction

Ole Hans and Benedikt Walter wrote a paper in two thousand twenty-five about choosing remote support for self-driving vehicles. Highly automated cars still meet situations they cannot handle. With no driver aboard, help must come from afar, either as remote driving, where a human steers directly, or remote assistance, where a human advises while the car drives itself.

Developers often pick by system design or by which option is simpler to build. The authors argue the real test is the match with the vehicle's defined operating domain and its use cases. They apply a known framework to describe that domain in a structured way. Their method weighs how well each support style complements the automated system in each scenario.

Catalog record

CREATOROle Hans, Benedikt Walter
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-14347

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Remote Assistance or Remote Driving: The Impact of Operational Design Domains on ADS-Supporting Systems Selection Ole Hans, Benedikt Walter Abstract High level Automated Driving Systems (ADS) can handle many situations, but they still encounter situations where human intervention is required. In systems where a physical driver is present in the vehicle, typically SAE Level 3 systems, this intervention is relatively straightforward and is handled by the in-vehicle driver. However, the complexity increases for Level 4 systems, where, in most cases, no physical driver remains in the vehicle. The two common industry solutions for this challenge are the integration of a remote support system, such as a Remote Driving System (RDS) or Remote Assistance System (RAS) . While it is clear that ADS will require one of these systems, it is less clear how the suitability of either system for a particular ADS application should be evaluated. Currently, the selection process often focuses on system architecture as well as its design and integration challenges. Furthermore, since many ADS developers choose to develop remote system solutions in-house, it is advantageous to select the simpler approach to streamline development and integration efforts. While these decision points are certainly relevant, this approach overlooks the most critical factors: the use cases and the complementarity of the ADS and the remote support system within the context of the Operational Design Design Domain (ODD) . This paper proposes a structured approach for selecting between RDS and RAS as an ADS support system, based on the defined ODD and use case analysis. To achieve this, the paper applies the PEGASUS framework to systematically describe and analyze the ODD. A structured framework is introduced to evaluate and select the most suitable remote support system for an ADS based on clearly defined criteria. I Introduction Automated Driving Systems (ADS) are widely regarded as a key technology for transforming the transportation sector, as they have the potential to enhance efficiency, safety, and driving comfort while reducing both the economic and environmental costs of mobility [ 1 ] . At the same time, ADS face significant challenges, particularly in urban environments where continuous interaction with other road users is required. A central issue in this context is the occurrence of so-called “stranded vehicle” situations, in which ADS encounter traffic scenarios that exceed their current technical capabilities and therefore necessitate human intervention. This challenge is further exacerbated by the system’s need to interpret and respond appropriately to the often unpredictable behavior of other traffic participants. As such, these limitations represent a challenge to the full automation of driving tasks and to the broader societal acceptance of such systems [ 2 ] . To resolve the potential situation whenever the ADS cannot perform the Dynamic Driving Tasks (DDT) anymore, these systems are equipped with a Minimum Risk Maneuver (MRM) capability that safely stops the vehicle to achieve a Minimum Risk Condition (MRC) . From that MRC, the ADS often is able to resume the DDT. However, in some situations that is not the case, which results in the fact that ADS still depends on supporting systems to handle specific scenarios beyond the capabilities of the ADS itself. Common solutions include the integration of Remote Driving System (RDS) and Remote Assistance Systems (RAS) , which, despite their similarities, are often chosen based on system architecture or short-term availability rather than a thorough evaluation of their suitability for particular ADS use cases. This approach frequently leads to suboptimal product-level outcomes, with certain essential use cases remaining unaddressed. Current robotaxi projects like Zoox or Waymo have chosen remote assistance solutions. It appears that this choice is made for two major reasons: • Firstly, RAS integrates smoothly into an existing ADS. The remote operator input merely replaces the perception input. The remaining ADS pipeline remains untouched, and the output is still generated by the ADS rather than the remote operator. • Second, most robotaxi developers have built their own remote systems. Therefore they naturally select the system that is less complex and has lower requirements for hardware, software, and performance. In contrast, companies like Vay Technology [ 3 ] and Roboauto [ 4 ] provide remote driving solutions that can be integrated in cars or trucks and allow to fully perform remote operation or support of an ADS system in a hybrid setup. The technology is more demanding than an RAS solution, therefore it is not by accident that there are companies solely focused on developing such systems. The main driver for selection between RAS and RDS therefore appears to be the integration effort and the overall complexity. This does not consider the actual reason why remote systems where needed in the first place. Supporting an ADS in a situation where it fails to perform the DDT due to a complex situation occurring in some scenario within the Operational Design Domain (ODD). Therefore, the authors argue that the primary decision point for the selection between RDS and RAS should be the ODD-related use case demands of an ADS system. While other factors might play a role, the main purpose of a remote technology is to enable an ADS to continue its DDT and therefore it must be evaluated which use cases for a given ODD cannot be performed by the ADS alone. This allows for an informed decision between RDS and RAS based on the technology capabilities and the resulting support for the different use cases. This paper presents a structured methodology for evaluating RAS and RDS within the context of the ODD. Based on the ODD definition part of the PEGASUS method [ 5 ] , the ODD is systematically analyzed to identify relevant use cases for these fallback systems, and the suitability of each system type is assessed accordingly. The results offer a comparative analysis of RAS and RDS as backup solutions for ADS, providing a foundation for informed decision-making and improved integration into ADS architectures. In the following paper, Section II provides an overview of the current state of research. Section III outlines the applied methodology. The core of the paper is presented in Section IV , where RAS and RDS are evaluated with respect to ODD-specific use cases. This comparison highlights the capabilities and limitations of each system and supports the selection of an appropriate fallback strategy in different ADS scenarios. The findings are discussed in detail in the result Section V and concluding a brief outlook on future work. II Literature Research A widespread concept for supporting unmanned vehicles is the concept of remote operation, such as remotely operated vehicles for Mars exploration or in the defense sector [ 6 ] . In the civilian sector, unmanned remote vehicles are often used for surveillance [ 7 ] or as fallback solutions for automated driving vehicles [ 8 ] . With regard to solving the challenges mentioned in Section I , the concept of remote operation is a solution to overcome the situations in which the ADS cannot handle a particular situation by calling for remote support. Remote operation can also intervene when automated driving is not expected to be available on certain sections of the route or at certain times of day. Once such situation occurs a predetermined takeover by remote operation takes

Read the complete text