Learning from experience is a fundamental capability of intelligent agents. Autonomous systems rely on sensors that provide data about the environment and internal situations to their perception systems for learning and inference mechanisms. These systems can also learn Self-Aware and Situation-Aware generative modules from these data to localize themselves and interact with the environment. In this paper, we propose a self-aware cognitive architecture capable to perform tasks where the interactions between the self-state of an agent and the surrounding environment are explicitly and dynamically represented. We specifically develop a Deep Learning (DL) based Self-Aware interaction model, empowered by learning from Multi-Modal Perception (MMP) and World Models using multi-sensory data in a novel Multi-Agent Self-Awareness Architecture (MASAA). Two sub-modules are developed, the Situation Model (SM) and the First-Person model (FPM), that address different and interrelated aspects of the World Model (WM). The MMP model, instead, aims at learning the mapping of different sensory perceptions into Exteroceptive (EI) and Proprioceptive (PI) latent information. The WM then uses the learned MMP model as experience to predict dynamic self-behaviors and interaction patterns within the experienced environment. WM and MMP Models are learned in a data-driven way, starting from the lower-dimensional odometry data used to guide the learning of higher-dimensional video data, thus generating coupled Generalized State Hierarchical Dynamic Bayesian Networks (GS-HDBNs). We test our model on KITTI, CARLA, and iCab datasets, achieving high performance and a low average localization error (RMSE) of 2.897%, when considering two interacting agents.

The combination of different sensory information to predict upcoming situations is an innate capability of intelligent beings. Consequently, various studies in the Artificial Intelligence field are currently being conducted to transfer this ability to artificial systems. Autonomous vehicles can particularly benefit from the combination of multi-modal information from the different sensors of the agent. This paper proposes a method for video-frame prediction that leverages odometric data. It can then serve as a basis for anomaly detection. A Dynamic Bayesian Network framework is adopted, combined with the use of Deep Learning methods to learn an appropriate latent space. First, a Markov Jump Particle Filter is built over the odometric data. This odometry model comprises a set of clusters. As a second step, the video model is learned. It is composed of a Kalman Variational Autoencoder modified to leverage the odometry clusters for focusing its learning attention on features related to the dynamic tasks that the vehicle is performing. We call the obtained overall model Cluster-Guided Kalman Variational Autoencoder. Evaluation is conducted using data from a car moving in a closed environment and leveraging a part of the University of Alcalá DriveSet dataset, where several drivers move in a normal and drowsy way along a secondary road.

This paper proposes a multi-modal situational inter-action model for collaborative agents by fusing multi-sensorial information in a Multi-Agent Hierarchical Dynamic Bayesian Network (MAH-DBN) framework. The proposed model is learned in a data-driven methodology to estimate the states of interacting agents only from video sequences. This can be regarded as a two-fold methodology for improving visual-based localization and interaction between autonomous agents. In the learning stage, the odometry model is used to drive the video learning model for a robust localization and interaction modeling. During the testing phase, the learned Multi-Agent Hierarchical DBN (MAH-DBN) model is used for the localization of collaborative agents only from video sequences by proposing an inference method called Multi-Agent Coupled Markov Jump Particle Filter (MAC-MJPF).

This paper proposes a method to localize a moving agent - such as an autonomous surveillance vehicle - inside a known environment using First Person Viewpoint video data. Anomalies w.r.t. expected vehicle motion and image content are extracted to guide the localization, signal when the localization results are not trustworthy and explain the reason for the failure. During the training phase, a Dynamic Bayesian Network model is learned, which couples positional and video data. To learn it, clustering is performed on the odometry data, and a modified Kalman Variational Autoencoder is built over the video data. During the testing phase, a Coupled Markov Jump Particle Filter leverages the learned Dynamic Bayesian Network to extract anomalies and to estimate the vehicle’s position, given only camera data. The proposed method is evaluated on two real-world datasets of a vehicle performing perimeter monitoring of a closed environment and of a shopping cart moving in a supermarket.

In the research area of human-robot interactions, the automatic estimation of the mass of a container manipulated by a person leveraging only visual information is a challenging task. The main challenges consist of occlusions, different filling materials and lighting conditions. The mass of an object constitutes key information for the robot to correctly regulate the force required to grasp the container. We propose a single RGB-D camera-based method to locate a manipulated container and estimate its empty mass i.e., independently of the presence of the content. The method first automatically selects a number of candidate containers based on the distance with the fixed frontal view, then averages the mass predictions of a lightweight model to provide the final estimation. Results on the CORSMAL Containers Manipulation dataset show that the proposed method estimates empty container mass obtaining a score of 71.08% under different lighting or filling conditions.

Anomaly detection constitutes a fundamental step in developing self-aware autonomous agents capable of continuously learning from new situations, as it enables to distinguish novel experiences from already encountered ones. This paper combines Dynamic Bayesian Networks (DBNs) and Neural Networks (NNs) and proposes a method for detecting anomalies in video data at different abstraction levels. We use a Variational Autoencoder (VAE) to reduce the dimensionality of video frames, and Optical Flows between subsequent images, generating a latent space that captures both visual and dynamical information and that is comparable to low-dimensional sensory data (e.g., positioning, steering angle). An Adapted Markov Jump Particle Filter is employed to predict the following frames and detect anomalies in video data. Our method’s evaluation is executed using different video data from a semi-autonomous vehicle performing different tasks in a closed environment. Tests on benchmark anomaly detection datasets have additionally been conducted.

This paper presents the outcomes of the PETS2021 challenge held in conjunction with AVSS2021 and sponsored by the EU FOLDOUT project. The challenge comprises the publication of a novel video surveillance dataset on through-foliage detection, the defined challenges addressing person detection and tracking in fragmented occlusion scenarios, and quantitative and qualitative performance evaluation of challenge results submitted by six worldwide participants. The results show that while several detection and tracking methods achieve overall good results, through-foliage detection and tracking remains a challenging task for surveillance systems especially as it serves as the input to behaviour (threat) recognition.

This paper proposes a method for performing future-frame prediction and anomaly detection on video data in a multi-modal framework based on Dynamic Bayesian Networks (DBNs). In particular, odometry data and video data from a moving vehicle are fused. A Markov Jump Particle Filter (MJPF) is learned on odometry data, and its features are used to aid the learning of a Kalman Variational Autoencoder (KVAE) on video data. Consequently, anomaly detection can be performed on video data using the learned model. We evaluate the proposed method using multi-modal data from a vehicle performing different tasks in a closed environment.

This paper proposes a method for performing continual learning of predictive models that facilitate the inference of future frames in video sequences. For a first given experience, an initial Variational Autoencoder, together with a set of fully connected neural networks are utilized to respectively learn the appearance of video frames and their dynamics at the latent space level. By employing an adapted Markov Jump Particle Filter, the proposed method recognizes new situations and integrates them as predictive models avoiding catastrophic forgetting of previously learned tasks. For evaluating the proposed method, this article uses video sequences from a vehicle that performs different tasks in a controlled environment.