Online data-driven anomaly detection in autonomous robots

Khalastchi, Eliahu; Kalech, Meir; Kaminka, Gal A.; Lin, Raz

doi:10.1007/s10115-014-0754-y

Cited by 77 publications

(35 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Generally, we agree with and target the requirements on monitoring systems introduced in Khalastchi et al [11], which state that these systems must be computationally light to prevent modifications and further failures in the system, must have a low false-positive rate and they should be able to detect context failures. These are failures that manifest in measurements which would be normal in a certain state, but are faulty in another one.…”

Section: Target Systems and Assumptionssupporting

confidence: 63%

“…For the evaluation, we have decided to use the F β score as the target metric with β = 0.1 to reflect the fact that a fault detection system with a high false-positive rate will most likely be ignored soon (cf. [11]). With this metric, all components in all test trials of the data set have been classified at the 1 Hz rate and the resulting boolean time series have been combined and the scores have been computed.…”

Section: Classificationmentioning

confidence: 99%

“…Khalastchi et al [11], in contrast, introduce an online fault detection approach which is purely data-driven and based on a sliding window approach to compare current measurements against the previous ones. It does not specifically address performance bugs and is mostly evaluated on sensor or hardware faults.…”

Section: Related Approachesmentioning

confidence: 99%

See 2 more Smart Citations

Autonomous fault detection for performance bugs in component-based robotic systems

Wienke

Wrede

2016

2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

Abstract-We present a novel fault detection method for application in component-based robotic systems. In contrast to existing work, our method specifically addresses faults in the software system of the robot using a data-driven methodology which exploits the inter-process communication of the system. This enables an application of the approach without expert knowledge or availability of complex software models. We specifically focus on performance bugs, which slowly degrade the performance of the system and are thereby harder to detect but also most valuable for automatic recovery. Using a data set recorded on a RoboCup@Home platform we demonstrate the performance and applicability of our method and analyze the types of faults that can be detected by the method.

show abstract

Section: Target Systems and Assumptionssupporting

confidence: 63%

Section: Classificationmentioning

confidence: 99%

See 1 more Smart Citation

Autonomous fault detection for performance bugs in component-based robotic systems

Wienke

Wrede

2016

2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

show abstract

“…Then, they identify abnormal values in each of the smaller-dimensional groups of variables using Mahalanobis distance [1]. Khalastchi et al [15] presented another approach, which is the improvement of their previous technique [1]. In this work they present an online data-driven anomaly detection approach, based on sliding window technique, which allows mining frequent patterns over data streams.…”

Section: A Anomaly Detection In Unmanned Vehiclesmentioning

confidence: 99%

UAV degradation identification for pilot notification using machine learning techniques

Manukyan

Olivares-Méndez

Bissyandé

et al. 2016

2016 IEEE 21st International Conference on Emerging Technologies and Factory Automation (ETFA)

View full text Add to dashboard Cite

Abstract-Unmanned Aerial Vehicles are currently investigated as an important sub-domain of robotics, a fast growing and truly multidisciplinary research field. UAVs are increasingly deployed in real-world settings for missions in dangerous environments or in environments which are challenging to access. Combined with autonomous flying capabilities, many new possibilities, but also challenges, open up. To overcome the challenge of early identification of degradation, machine learning based on flight features is a promising direction. Existing approaches build classifiers that consider their features to be correlated. This prevents a fine-grained detection of degradation for the different hardware components. This work presents an approach where the data is considered uncorrelated and, using machine learning techniques, allows the precise identification of UAV's damages.

show abstract

“…In general, fault detection algorithms can be classified into three types: model-based, knowledge-based, and data-driven-based approaches [ 4 ]. For the classic model-based fault detection and diagnosis, the research could be classified into two distinct and parallel communities [ 5 ], i.e., the Fault Detection and Isolation (FDI) community and Diagnosis (DX) community.…”

Section: Introductionmentioning

confidence: 99%

A Novel Online Data-Driven Algorithm for Detecting UAV Navigation Sensor Faults

Sun

Cheng

Wang

et al. 2017

Sensors

View full text Add to dashboard Cite

The use of Unmanned Aerial Vehicles (UAVs) has increased significantly in recent years. On-board integrated navigation sensors are a key component of UAVs’ flight control systems and are essential for flight safety. In order to ensure flight safety, timely and effective navigation sensor fault detection capability is required. In this paper, a novel data-driven Adaptive Neuron Fuzzy Inference System (ANFIS)-based approach is presented for the detection of on-board navigation sensor faults in UAVs. Contrary to the classic UAV sensor fault detection algorithms, based on predefined or modelled faults, the proposed algorithm combines an online data training mechanism with the ANFIS-based decision system. The main advantages of this algorithm are that it allows real-time model-free residual analysis from Kalman Filter (KF) estimates and the ANFIS to build a reliable fault detection system. In addition, it allows fast and accurate detection of faults, which makes it suitable for real-time applications. Experimental results have demonstrated the effectiveness of the proposed fault detection method in terms of accuracy and misdetection rate.

show abstract

Online data-driven anomaly detection in autonomous robots

Cited by 77 publications

References 24 publications

Autonomous fault detection for performance bugs in component-based robotic systems

Autonomous fault detection for performance bugs in component-based robotic systems

UAV degradation identification for pilot notification using machine learning techniques

A Novel Online Data-Driven Algorithm for Detecting UAV Navigation Sensor Faults

Contact Info

Product

Resources

About