Subsumption Architecture Research Articles

General Intelligent Dataset Description Method and Application

With the development of the Internet, the growth rate of big data has exceeded the ability of human beings to obtain information from data. In the near future, faced with the sea of data, seemingly logically correct algorithms may come to the opposite conclusion, which will be a “data disaster” era. Therefore, letting data “speak” and enable data intelligence is an important problem that must be solved in the field of big data research. In this paper, we propose a “data intelligence” framework based on Brooks’s Subsumption Architecture for robotics. In this framework, we define a new dataset description method and try to encapsulate the dataset described by this method into a software container. Then, we set up the corresponding operations on the dataset container. Finally, we identify an interaction model for a “dataset agent” based on academic work on agents in the field of artificial intelligence. And we designed and implemented some functions of an intelligent document manager, which proves that our design ideas and methods are feasible.

Roles for cerebellum and subsumption architecture in central pattern generation.

Within vertebrates, central pattern generators drive rhythmical behaviours, such as locomotion and ventilation. Their pattern generation is also influenced by sensory input and various forms of neuromodulation. These capabilities arose early in vertebrate evolution, preceding the evolution of the cerebellum in jawed vertebrates. This later evolution of the cerebellum is suggestive of subsumption architecture that adds functionality to a pre-existing network. From a central-pattern-generator perspective, what additional functionality might the cerebellum provide? The suggestion is that the adaptive filter capabilities of the cerebellum may be able to use error learning to appropriately repurpose pattern output. Examples may include head and eye stabilization during locomotion, song learning, and context-dependent alternation between learnt motor-control sequences.

Decentralised aerial swarm for adaptive and energy efficient transport of unknown loads

• We present a behaviour based subsumption architecture that enables a swarm’s topology to adapt to a load's mass distribution. • The decentralized self-organized swarm can perform a cooperative load transportation task without any explicit communication between members. • Our architecture has the potential to be scalable to a large number of agents. • This is significant in aerial swarm cooperative transport systems because it leads to even energy distribution among swarm members when compared to a uniform swarm topology. • It also fosters robustness of the load-suspended aerial robot system. Cooperative transport by a swarm of Quadcopters offers more flexibility and performance when carrying loads that are complex in structural profile and mass. Ensuring that team members of the swarm are optimally placed on these loads as well as able to resist disturbances from the environment during transport are current research challenges. In this paper, we present a decentralized behaviour based subsumption architecture for enabling a swarm of Quadcopters to explore an unfamiliar area, find a load and transport it to a target location cooperatively. In the architecture, three behaviours were used: an obstacle avoidance behaviour to avoid collisions with objects in the environment, a flocking behaviour to ensure swarm structure and a bacterium behaviour for exploration of the environment and to adapt to the mass profile of various detected loads. By adapting to the mass profile of a detected load, we show that our architecture ensures even energy distribution among Quadcopters while achieving robustness to disturbances from the environment. Our results show that a mass adapting swarm is able to conserve energy during payload transportation when compared to a swarm that does not adapt to a load’s profile. Furthermore, we do not use explicit communication between team members but instead rely on data from visual sensors attached to the Quadcopters. We experiment with simulations in a physics informed robot simulator called CoppeliaSim and demonstrate the effectiveness of our architecture when utilized for cooperative transport of irregular loads.

Agent Architecture in Social Games — The Implementation of Subsumption Architecture in Diplomacy

Social games are challenging for AI research because they involve not only the mechanism of game (taking actions, planning, etc), but also the social aspects including communicating, cooperating and reaching agreements. Some social games also require cunning, duplicity, or bad faith. In this paper we present the implementation of a prototype that utilizes the properties of the subsumption architecture to provide an interesting computer opponent in a social board game — Diplomacy. The evaluation result indicates that the subsumption architecture is appropriate for social games.

An AI System for Large Open Virtual World

In recent years, computer games have reached unprecedented level of graphical fidelity to the real world. As the non-player characters (NPCs) in the game world look more and more realistic, players expect them to manifest believable behavior as well. This is accented especially in games that feature large open worlds, which players may explore freely and it is thus not possible to explicitly account for all possible player interactions. In this paper we focus mainly on ambient AI - the logic behind day to day behaviors of NPCs as they sleep, work and entertain themselves in the virtual world. In this context, it is of great importance to build a system that handles many NPCs (up to several hundreds) quickly. In this paper we report on an implementation of a particular AI system that was approved for deployment in an upcoming high-budget game. The system features a hierarchy of control similar to the subsumption architecture and a visual agent-based language inspired by behavior trees. We describe the challenges involved in building such a system and specific design decisions we have made that let us achieve a level of behavioral fidelity unmatched by existing games. Finally we evaluate the performance of the system in a realistic setting.

A novel plume tracking method in partial 3D diffusive environments using multi-sensor fusion

In recent years, dangerous gas leakage has led to serious consequences in social security. Pervious strategies for locating the odor sources were carried out in 2D environments or the gas sensors were installed at a fixed height, ignoring the characteristics of plume distribution in 3D environments. In some cases, strategies developed in 2D environments may be completely ineffective when be applied to 3D environments. To enhance the success rate and rapidity, a novel method of odor source location in partial 3D diffusive environments based on multi-sensor fusion is proposed. Considering the suspected source existed and weak wind environments, machine olfactory and vision information are both employed to track the plumes and identify its source. To enhance the speed of plume searching and the reliability of plume tracking in the early stages, an autonomous mobile robot (AMR) simulates the social mechanism and hunting behaviors of the gray wolf population and tracks the plumes. A technical strategy is also designed to enable the AMR to complete the task of gas source location successfully and the subsumption architecture is adopted to define and arbitrate behavior priorities to coordinate different behaviors in this paper. The on-site test results show that the average positioning error is 0.13 m, the average running time is 147.7 s and the average distance traveled is 21.35 m. The results show that the proposed method is competent to accomplish the task of leakage gas source localization in partial 3D diffusive environments.

Simulation of Pulse-Echo Radar for Vehicle Control and SLAM.

Pulse-echo sensing is the driving principle behind biological echolocation as well as biologically-inspired sonar and radar sensors. In biological echolocation, a single emitter sends a self-generated pulse into the environment which reflects off objects. A fraction of these reflections are captured by two receivers as echoes, from which information about the objects, such as their position in 3D space, can be deduced by means of timing, intensity and spectral analysis. This is opposed to frequency-modulated continuous-wave radar, which analyses the shift in frequency of the returning signal to determine distance, and requires an array of antenna to obtain directional information. In this work, we present a novel simulator which can generate synthetic pulse-echo measurements for a simulated sensor in a virtual environment. The simulation is implemented by replicating the relevant physical processes underlying the pulse-echo sensing modality, while achieving high performance at update rates above 50 . The system is built to perform design space exploration of sensor hardware and software, with the goals of rapid prototyping and preliminary safety testing in mind. We demonstrate the validity of the simulator by replicating real-world experiments from previous work. In the first case, a subsumption architecture vehicle controller is set to navigate an unknown environment using the virtual sensor. We see the same trajectory pattern emerge in the simulated environment rebuilt from the real experiment, as well as similar activation times for the high-priority behaviors (±1.9%), and low-priority behaviors (±0.2%). In a second experiment, the simulated signals are used as input to a biologically-inspired direct simultaneous mapping and localization (SLAM) algorithm. Using only path integration, 83% of the positional errors are larger than 10 , while for the SLAM algorithm 95% of the errors are smaller than . Additionally, we perform design space exploration using the simulator. By creating a synthetic radiation pattern with increased spatiospectral variance, we are able to reduce the average localization error of the system by 11%. From these results, we conclude that the simulation is sufficiently accurate to be of use in developing vehicle controllers and SLAM algorithms for pulse-echo radar sensors.

Adaptive Sampling of Surface Fronts in the Arctic Using an Autonomous Underwater Vehicle

Fronts between Arctic- and Atlantic-origin waters are characterized by strong lateral gradients in temperature and salinity. Ocean processes associated with fronts are complex with considerable space and time variability. Therefore, resolving the processes in frontal zones by observation is challenging but important for understanding the associated physical–biological interactions and their impact on the marine ecosystem. The use of autonomous robotic vehicles and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> data-driven sampling can help improve and augment the traditional sampling practices, such as ships and profiling instruments. Here, we present the development and results of using an autonomous agent for detection and sampling of an Arctic front, integrated on board an autonomous underwater vehicle. The agent is based on a subsumption architecture implemented as behaviors in a finite-state machine. Once a front is detected, the front tracking behavior uses observations to continuously adapt the path of the vehicle to perform transects across the front interface. Following successful sea trials in the Trondheimsfjord, the front-tracking agent was deployed to perform a full-scale mission near 82 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{\circ}$</tex-math></inline-formula> N north of Svalbard, close to the sea ice edge. The agent was able to detect and track an Arctic frontal feature, performing a total of six crossings while collecting vertical profiles in the upper 90 m of the water column. Measurements yield a detailed volumetric description of the frontal feature with high resolution along the frontal zone, augmenting ship-based sampling that was run in parallel.

Development of a Novel Robust Control Method for Formation of Heterogeneous Multiple Mobile Robots With Autonomous Docking Capability

Multiple mobile robots in formation are often required to dock to each other to overcome the limitations, such as battery failure, transportation capacity, and maneuverability on rough terrains; however, it is challenging to design a single controller that navigates the robots to dock to each other, maintains the other robots in formation, and is applicable to both docked and nondocked robots, while it is also robust to uncertainties and disturbances. This article proposes a novel robust subsumption architecture for nonholonomic mobile robots in formation with docking capability. In addition to docking, the robots, i.e., all the nondocked robots and the front-docked robots, maintain a formation that can also be switched automatically to other configurations when necessary and avoid collisions with other robots and dynamic obstacles. The proposed subsumption control architecture takes into account each follower’s desired goal as well as its docking condition to synthesize a control law as a velocity control signal that is then used to determine the robust input torque for each follower using the robots’ dynamics. The Lyapunov stability of the controller is also proved. We also develop strategies for efficient centralized motion planning of the followers to achieve various goals, e.g., formation keeping/switching, docking, and collision avoidance. The effectiveness of our proposed methodology was verified in simulations as well as implementations on a virtual robot environment. Note to Practitioners —Multiple mobile robots, especially when operating as a formation, are able to perform tasks that are beyond the capabilities of individual robots. Existing formation control approaches neglect some realistic limitations of mobile robots, such as battery failure, limited transportation capacity, and maneuverability, to name a few. This article was motivated by these realistic limitations of mobile robots when operating in formation, and it suggests a new approach for navigation of such robots by docking some (or all) of these robots to each other and pursue a variety of goals. The goal includes autonomous docking, formation keeping/switching, and collision avoidance in dynamic environments. We include robot dynamics and system uncertainties in our algorithm and provide a robust control methodology. Therefore, the developed methodologies in this article can be adopted in real applications that require robots to be supplied with sufficient battery or having a large payload capacity, e.g., agricultural robotics.

Sequential Localizing and Mapping: A Navigation Strategy via Enhanced Subsumption Architecture.

In this paper, we present a navigation strategy exclusively designed for social robots with limited sensors for applications in homes. The overall system integrates a reactive design based on subsumption architecture and a knowledge system with learning capabilities. The component of the system includes several modules, such as doorway detection and room localization via convolutional neural network (CNN), avoiding obstacles via reinforcement learning, passing the doorway via Canny edge’s detection, building an abstract map called a Directional Semantic Topological Map (DST-Map) within the knowledge system, and other predefined layers within the subsumption architecture. The individual modules and the overall system are evaluated in a virtual environment using Webots simulator.

Reactive control scheme based on fuzzy behaviors for mobile robot

This paper presents the design and implementation of a behavior-based control scheme. The construction of the set of behaviors is based on the use of fuzzy logic as a means for materializing the designer’s knowledge into the behaviors. The behavior set was established by left and right wall following and obstacle avoidance. These three behaviors were programmed and coordinated by a subsumption architecture or behavioral inhibition. Behavior simulations and coordination scheme design were tested by means of real experiments using a mobile robotic platform. Finally, the results are presented, where the control actions are executed by the robotic system achieving a secure navigation.

Preliminary Development of Subsumption Architecture Control for Automated Guided Vehicle

Subsumption control architecture is an control architecture based on parallel system. Input of information of sensors is directly connected through modules in the control system, and further the decision making is connected to actuators. Automated Guided Vehicle or AGV is an automated component within integrated manufacturing system. In this article, this control architecture will be designed and implemented to an AGV. Commands are designed based on Object-Oriented technology. The commands are arranged in subsumption, where a command higher subsumed other command of its lower level. GPFO (Greater Priority First Out) technique is implemed for executing the commands by using multi-threading. Experimentation is performed to have the characteristics of commands being executed. This work introduce our effort to design an operating system for an AGV.

A Biomimetic Radar System for Autonomous Navigation

This paper presents a novel biomimetic radar sensor for autonomous navigation. To accomplish this, we have drawn inspiration from the sensory mechanisms present in an echolocating mammal, the common big-eared bat ( Micronycteris microtis ). We demonstrate the correspondence in both the hardware, system model, and signal processing. To validate the performance of the sensor, we have developed a complementary control system based on subsumption architecture, which allows the system to autonomously navigate unknown environments. This architecture consist of separate behaviors with different levels of complexity, which are combined to produce the overall functionality of the system. We describe each behavior separately and examine their performance in real-world navigation experiments. For this purpose, the system is placed in two distinct office environments with the goal of achieving smooth and stable trajectories. Here, we can observe noticeable improvements when employing high-level behaviors. Furthermore, we utilize the data collected during the navigation experiments to perform simultaneous localization and mapping, using an algorithm developed in our earlier work. These results show a substantial improvement over the odometry. We attribute this to the fact that the system traverses stable and repetitive paths, which facilitates place recognition.

On Reliable Neural Network Sensorimotor Control in Autonomous Vehicles

This paper deals with (deep) neural network implementations of sensorimotor control for automated driving. We show how to construct complex behaviors by re-using elementary neural network building blocks that can be trained and tested extensively; one of our goals is to mitigate the “black box” and verifiability issues that affect end-to-end trained networks. By structuring complex behaviors within a subsumption architecture, we retain the ability to learn (mostly at motor primitives level) with the ability to create complex behaviors by subsuming the (well-known) learned elementary perception-action loops. The learning process itself is simplified, since the agent needs only to learn elementary behaviors. At the same time, the structure imposed with the subsumption architecture ensures that the agent behaves in predictable ways (e.g., treating all obstacles uniformly). We demonstrate these ideas for longitudinal obstacle avoidance behavior, but the proposed approach can also be adapted to other situations.

Proposal of Layered Control System Based on Sensor Information

The current control system for the robot is valid if its environment is precisely known, however, it is invalid for the unknown environment such as space investigation and disaster area. Although the subsumption architecture is famous that it is robust to the environmental change, it has some problems: the layer definition is unclear and the way to resolve the confliction among layers is unclear. We propose to use sensor information as layer definition. As a result, confliction among layers is basically not generated. The proposed method is applied to a mobile robot and numerical simulation results are shown.

A cooperative architecture for swarm robotic based on dynamic fuzzy cognitive maps

This work presents a cooperative architecture for navigation of a swarm of robots based on Dynamic Fuzzy Cognitive Maps (DFCM). This architecture is used to develop homogeneous autonomous robots without any type of global controllers. The developed autonomous navigation system has skills for learning, self-adaptation, behavior management and cooperative data share. We adopt a multi-agent approach, inspired by Brooks subsumption architecture, which allows hierarchical management of actions and parallel processing. Reinforcement learning is used to self-tune the navigation system parameters allowing the DFCM model to be self-adaptive. Two strategies are analyzed for data and experience exchange between agents (robots). The first one is based on a navigation memory sharing and the other is a bio inspired by ant's behavior strategy. Results are presented with both strategies and comparisons of their performance are carried out in a virtual environment.

How Behavior Trees Modularize Hybrid Control Systems and Generalize Sequential Behavior Compositions, the Subsumption Architecture, and Decision Trees

Behavior trees (BTs) are a way of organizing the switching structure of a hybrid dynamical system (HDS), which was originally introduced in the computer game programming community. In this paper, we analyze how the BT representation increases the modularity of an HDS and how key system properties are preserved over compositions of such systems, in terms of combining two BTs into a larger one. We also show how BTs can be seen as a generalization of sequential behavior compositions, the subsumption architecture, and decisions trees. These three tools are powerful but quite different, and the fact that they are unified in a natural way in BTs might be a reason for their popularity in the gaming community. We conclude the paper by giving a set of examples illustrating how the proposed analysis tools can be applied to robot control BTs.

The cognitive systems toolkit and the CST reference cognitive architecture

Abstract In this paper, we introduce the Cognitive Systems Toolkit (CST) and its underlying CST Reference Cognitive Architecture. CST is a general toolkit for the construction of cognitive architectures, which relies on a set of concepts which are familiar to many other cognitive architecture and constitute CST’s core. This core is general enough such that problem-specific cognitive architectures might be generated by using CST. At the same time, CST specify a more general Reference Cognitive Architecture, which is a reference model for the construction of application specific cognitive architectures. Differently from other Cognitive Architectures available in the literature, which are computational frameworks, CST is a toolkit, which means more flexibility in choosing specific techniques and algorithms for setting up the application architecture. After presenting the architecture core, we develop an illustrative example showing how CST can be used to implement a simple subsumption architecture in a robotic application.

Subsumption architecture applied to flight control using composite rotations

In this paper the subsumption theory is applied to flight control through composite rotations where multiple tasks can be defined as simple rotations. The tasks can then be arranged as a hierarchy, where the primary task is always fully pursued, and conflicting lower level tasks are removed by the primary rotation. The concept is applied to a group of uavs that move through an urban terrain while avoiding collisions with the ground, the buildings and other uavs as they track a desired waypoint.

On Curve Negotiation: From Driver Support to Automation

This paper describes a curve negotiation “behavior” that can be used—within subsumption architectures—to produce artificial agents with the ability to negotiate curves in a humanlike way. This may be used to implement functions spanning different levels of automation, from assistance (curve warning) to automated (curve speed control). This paper gives the following: 1) a summary of related works and of the subsumption architecture conceptual framework; 2) a detailed description of the function within this framework; 3) experimental data for validation and tuning derived from user tests; 4) guidelines on integration of the function within advanced driver assistance systems with different automation levels, with examples; and 5) a comparison with experimental data of the human curve speed choice models in the state of the art.