Incident: Volvo's Self-Driving Car Fails to Detect Kangaroos Due to Hopping.

Published Date: 2017-06-30

Postmortem Analysis
Timeline 1. The software failure incident where Volvo's self-driving car was unable to detect kangaroos due to their hopping movement happened during the testing phase in Australia [60129]. 2. The incident occurred before the article was published on June 30, 2017. 3. Therefore, the software failure incident with kangaroo detection likely happened in the first half of 2017.
System 1. Volvo's "Large Animal Detection system" failed to detect kangaroos due to their unique hopping movement [60129].
Responsible Organization 1. Volvo's self-driving car software system [60129]
Impacted Organization 1. Volvo (Volvo Australia) [60129] 2. Robert Bosch Australia [60129]
Software Causes 1. The software cause of the failure incident was the inability of Volvo's self-driving car's "Large Animal Detection system" to adjust to the unique movement of kangaroos due to their hopping, which confounded the system [60129].
Non-software Causes 1. The unique method of movement of kangaroos, specifically their hopping, confounded the self-driving car's systems [60129]. 2. The car's object detection systems used the ground as a reference point, making it difficult to judge the distance of kangaroos when they were in the air versus when they landed [60129].
Impacts 1. The software failure incident where Volvo's self-driving car was unable to detect kangaroos due to their hopping movement could potentially lead to accidents involving kangaroos, as kangaroos are responsible for about 90% of collisions between vehicles and animals in Australia [60129]. 2. This incident highlighted a limitation in Volvo's Large Animal Detection system, which could impact the safety and effectiveness of their autonomous driving technology, especially in regions where kangaroos are prevalent [60129].
Preventions 1. Conducting thorough testing in diverse environments: The software failure incident where Volvo's self-driving car was unable to detect kangaroos could have been prevented by conducting more extensive testing in various environments, including regions with unique animal behaviors like Australia [60129]. 2. Implementing adaptive object detection systems: To prevent such incidents in the future, carmakers like Volvo could consider implementing adaptive object detection systems that can adjust to different types of movements, such as the hopping of kangaroos, rather than solely relying on the ground as a reference point [60129].
Fixes 1. Implementing a software update to adjust the object detection system to account for the unique movement of kangaroos, such as their hopping pattern. This update could involve refining the algorithms used for distance estimation and object recognition [60129].
References 1. Volvo Australia's managing director, Kevin McCann [Article 60129] 2. Volvo's Australian technical manager, David Pickett [Article 60129] 3. Spokeswoman for Robert Bosch Australia, Amy Kaa [Article 60129]

Software Taxonomy of Faults

Category Option Rationale
Recurring unknown (a) The software failure incident related to Volvo's self-driving car being unable to detect kangaroos due to their hopping movement is specific to Volvo. There is no mention in the article of a similar incident happening before within the same organization or with its products and services. (b) The article does not mention any similar incident happening at other organizations or with their products and services.
Phase (Design/Operation) design, operation (a) The software failure incident related to the design phase can be seen in the article where Volvo's self-driving car was unable to detect kangaroos due to the unique movement of the animals. The article mentions that the company's "Large Animal Detection system" was initially designed to identify and avoid animals like deer, elk, and caribou but faced challenges in adjusting to the kangaroo's hopping movement. This design limitation was discovered during testing in Australia, highlighting a failure in the design phase of the autonomous driving technology [60129]. (b) The software failure incident related to the operation phase can be inferred from the article where Volvo's Australian technical manager, David Pickett, explained that the troubles with detecting kangaroos arose because their cars' object detection systems used the ground as a reference point. This operational limitation led to difficulties in judging the distance of kangaroos, especially when they were in the air hopping. Therefore, the operation of the system, specifically how it interacted with the environment and detected objects, contributed to the failure in detecting kangaroos effectively [60129].
Boundary (Internal/External) within_system (a) within_system: The software failure incident related to Volvo's self-driving car not being able to detect kangaroos due to their hopping confounding the system is a result of factors originating from within the system. The article mentions that the trouble arose because the cars' object detection systems used the ground as a reference point, making it difficult to judge the distance of kangaroos when they are hopping [60129]. This indicates that the issue is internal to the system's design and detection mechanisms.
Nature (Human/Non-human) non-human_actions (a) The software failure incident in this case is related to non-human actions. Volvo's self-driving car's inability to detect kangaroos was due to the unique movement of kangaroos confounding its systems. The managing director of Volvo Australia mentioned that the troubles arose because their cars' object detection systems used the ground as a reference point, making it difficult to judge the distance of kangaroos when they were hopping in the air [60129]. The software failure was a result of the inherent nature of kangaroos' movement, which was not intentionally introduced by humans but rather a challenge faced during the development and testing of driverless technology.
Dimension (Hardware/Software) hardware (a) The software failure incident related to hardware: - The article mentions that Volvo's self-driving car had trouble detecting kangaroos due to the unique movement of the animals, specifically their hopping behavior. This issue was attributed to the cars' object detection systems using the ground as a reference point, which made it difficult to judge the distance of kangaroos when they were in the air versus on the ground [60129]. (b) The software failure incident related to software: - The software failure incident in this case was not directly attributed to software issues but rather to the challenge of adapting the software to recognize and adjust to the unique movement patterns of kangaroos. The article highlights that Volvo was working on developing technology to recognize kangaroos as part of the ongoing development of autonomous cars [60129].
Objective (Malicious/Non-malicious) non-malicious (a) The software failure incident related to the inability of Volvo's self-driving car to detect kangaroos is non-malicious. The failure was due to the unique movement of kangaroos confounding the car's "Large Animal Detection system" rather than any malicious intent by humans [60129]. The company's managing director mentioned that the discovery of this issue was part of the development and testing of driverless technology, indicating a non-malicious nature of the incident. Additionally, the article highlights the efforts made by Volvo to gather data on how animals move and behave to improve the system's understanding, further supporting the non-malicious nature of the failure.
Intent (Poor/Accidental Decisions) unknown (a) The software failure incident related to Volvo's self-driving car being unable to detect kangaroos due to their hopping movement does not seem to be a result of poor decisions. Instead, it appears to be a challenge faced during the development and testing of driverless technology. The company's managing director mentioned that this discovery was part of the ongoing development process for autonomous cars and that gathering data on how animals move and behave is essential for the computers to understand it better [60129]. (b) The incident does not indicate that the failure was due to accidental decisions or mistakes. It is portrayed as a technical challenge that arose during the testing phase of the self-driving car technology, specifically in adapting the object detection system to recognize the unique movement of kangaroos. The article emphasizes the continuous development of autonomous cars and the need to address various environmental factors and conditions to enhance the technology [60129].
Capability (Incompetence/Accidental) accidental (a) The software failure incident in this case does not seem to be related to development incompetence. The article mentions that Volvo's self-driving car was unable to detect kangaroos due to the unique movement of the animals, specifically their hopping, which confounded the car's systems. This issue was discovered during testing in Australia, and Volvo's engineers were gathering data on how the animals move and behave to improve the system [60129]. (b) The software failure incident appears to be accidental. Volvo's Australian technical manager mentioned that the trouble with detecting kangaroos arose because their cars' object detection systems used the ground as a reference point, making it difficult to judge the distance of kangaroos when they were in the air versus on the ground. This accidental factor of using the ground as a reference point led to the difficulty in detecting kangaroos due to their hopping movement [60129].
Duration temporary The software failure incident described in the articles can be categorized as a temporary failure. The incident where Volvo's self-driving car was unable to detect kangaroos due to their unique hopping movement was a temporary issue caused by specific circumstances related to the kangaroo's behavior [60129]. The company acknowledged this issue as part of the ongoing development and testing of their driverless technology, indicating that it was a specific challenge that needed to be addressed through data gathering and adjustments to the detection system. The software failure was not permanent but rather a temporary setback that required further refinement and development to overcome.
Behaviour omission, other (a) crash: The software failure incident described in the article is not related to a crash where the system loses state and does not perform any of its intended functions. Instead, it focuses on the inability of Volvo's self-driving car to detect kangaroos due to their unique movement patterns [60129]. (b) omission: The issue with Volvo's self-driving car not being able to detect kangaroos can be categorized as an omission failure. The system is omitting to perform its intended function of accurately detecting and responding to kangaroos due to the hopping movement confusing its detection systems [60129]. (c) timing: There is no indication in the article that the software failure incident is related to timing issues where the system performs its intended functions too late or too early. The focus is on the system's inability to detect kangaroos accurately [60129]. (d) value: The software failure incident is not attributed to the system performing its intended functions incorrectly in terms of the value provided. Instead, it is about the system's inability to detect kangaroos effectively due to their unique movement patterns [60129]. (e) byzantine: The software failure incident does not exhibit characteristics of a byzantine failure where the system behaves erroneously with inconsistent responses and interactions. The issue with kangaroo detection is more about the system's limitations in understanding the animal's movements [60129]. (f) other: The behavior of the software failure incident can be categorized as a limitation in the system's object detection capabilities when it comes to unique animal movements like kangaroos' hopping. This can be seen as a specific type of failure related to environmental adaptation and animal behavior recognition, which is not explicitly covered in the provided options [60129].

IoT System Layer

Layer Option Rationale
Perception None None
Communication None None
Application None None

Other Details

Category Option Rationale
Consequence delay, non-human, theoretical_consequence (a) death: People lost their lives due to the software failure - There is no mention of people losing their lives due to the software failure incident in the provided article [60129]. (b) harm: People were physically harmed due to the software failure - There is no mention of people being physically harmed due to the software failure incident in the provided article [60129]. (c) basic: People's access to food or shelter was impacted because of the software failure - There is no mention of people's access to food or shelter being impacted due to the software failure incident in the provided article [60129]. (d) property: People's material goods, money, or data was impacted due to the software failure - There is no mention of people's material goods, money, or data being impacted due to the software failure incident in the provided article [60129]. (e) delay: People had to postpone an activity due to the software failure - The software failure incident related to Volvo's self-driving car not being able to detect kangaroos due to their unique movement pattern could potentially cause delays in the development and testing of driverless technology [60129]. (f) non-human: Non-human entities were impacted due to the software failure - The software failure incident impacted the detection system's ability to recognize kangaroos due to their hopping movement, which is a unique challenge for the system [60129]. (g) no_consequence: There were no real observed consequences of the software failure - The software failure incident did have consequences related to the detection system's inability to recognize kangaroos due to their hopping movement [60129]. (h) theoretical_consequence: There were potential consequences discussed of the software failure that did not occur - The article discusses the potential consequences of kangaroos causing accidents in Australia due to the detection system's inability to recognize them, but it does not mention any actual accidents or incidents occurring as a result of this software failure [60129]. (i) other: Was there consequence(s) of the software failure not described in the (a to h) options? What is the other consequence(s)? - There are no other consequences mentioned in the article beyond the impact on the detection system's ability to recognize kangaroos due to their unique movement pattern [60129].
Domain transportation, unknown (a) The failed system was intended to support the transportation industry. The incident involved Volvo's self-driving car, which was unable to detect kangaroos due to their unique hopping movement, causing a challenge for the car's object detection systems [60129]. The article specifically mentions Volvo's development and testing of driverless technology for their autonomous cars, which are aimed at revolutionizing transportation by introducing self-driving capabilities [60129]. (b) The failed system was intended to support the transportation industry. Volvo's self-driving car, which encountered the issue of not being able to detect kangaroos effectively, is a part of the company's efforts to develop autonomous driving technology for transportation purposes [60129]. (c) Unknown (d) Unknown (e) Unknown (f) Unknown (g) Unknown (h) Unknown (i) Unknown (j) Unknown (k) Unknown (l) Unknown (m) Unknown

Sources

Back to List