Welcome to ARRI's Distributed Intelligence & Autonomy Lab (DIAL) webpage.

The mission of the research group is to study distributed autonomous systems for environmental monitoring applications in poorly structured environments. Thanks to powerful lab capabilities (including ground and aerial mobile platforms and sensor motes), a wide variety of experiments are currently being performed to test innovative control algorithms in real-world scenarios.

 

Projects (please check videos in the multimedia section)

Discrete event coordination
The key research issue is to endow the MSN with the sufficient intelligence to automatically react to stimuli of external environments according to a predefined set of cooperation rules.
We use a matrix-based discrete event controller as a single framework for combined design of task planning, resource assignment, and deadlock avoidance of a mobile sensor network with multiple missions. This control architecture also provides an intuitive tool for easily programming the mission goals and priorities, which is a major concern in MSN if external conditions change or more information is available to a human operator. (View video: here)

Dynamic resource allocation for Air/Ground Mobile sensor network with localization
In unstructured environments, dynamic resource assignment is required for effective cooperation of robot teams. In some scenarios, robots are in charge of executing multiple missions simultaneously. This creates risks of deadlock due to the presence of shared resources among various missions. The main contribution of this research is the development of a novel approach that combines the one step look-ahead deadlock avoidance policy with dynamic resource assignment. The dynamic resource assignment is achieved using greedy resource assignment for multi-mission robot teams in the framework of a matrix based discrete event controller.
 

Collaborative exploration, map building, and trust
In this research, we explore the topics of path building with exploration, coordination of tasks and secure environment for robot team cooperation. Mobile wireless sensor networks are more vulnerable to security attacks than wired networks due to the broadcast nature of the transmission medium. Also, they have additional vulnerability as the mobile nodes are often placed in hostile environments where they cannot be physically protected. Since the base station is the gateway for the nodes to communicate with each other, compromising the base station can render the entire WSN useless. During the creation of WSN, each node is given a master key which is shared with the base station. All other keys are derived from this key. Mobile nodes can move in and out of the network, and hence have to be authenticated when they reenter the network. This gives rise to task planning where some mobile nodes have to intercept the incoming node and verify the master key. Such task planning can be easily done using a supervisory discrete event controller. For example, the test bed at ARRI consists of mobile robots (Acroname® Garcia) and a maze. There is a sentry robot which patrols the maze. This robot is localized using reference MIT Cricket ultrasound sensors. The sentry robot is programmed to perform obstacle avoidance, so that it can travel from the entry of the maze to the exit. In its journey, its path is recorded at the base station and sent back to the sentry robot. When another robot enters the grid (one which does not have obstacle avoidance), an event is triggered and the discrete event controller jumps into action. The guard robot intercepts the new mobile node, and pings for its network key. If the key is invalid, the new node is not allowed to enter the maze. If the key is valid, the sentry robot sends the path information to the new robot and allows it to pass through the maze.

Condition Based Maintenance
A new application architecture is designed for continuous, real-time, distributed wireless sensor networks. We develop a wireless sensor network for machinery condition-based maintenance (CBM) using commercially available products. We develop a hardware platform, networking architecture, and medium access communication protocol. We implement a single-hop sensor network to facilitate real-time monitoring and extensive data processing for machine monitoring. A new radio battery consumption model is presented and the given battery consumption equation is used to select the most suitable topology and design an energy efficient communication protocol for wireless sensor network. A new streamlined matrix formulation is developed which allows the base station to compute best periodic sleep times for all the nodes in the network. We combine scheduling and contention to design a hybrid MAC protocol, which achieves 100 % collision avoidance by using our modified RTS – CTS contention mechanism known as UC-TDMA protocol. A LabVIEW graphical user interface is described that allows for signal processing, including FFT, various moments, and kurtosis. A wireless CBM sensor network implementation on a Heating & Air Conditioning Plant is presented as a case study.

Self-Localization of Sensor Nodes
The main idea is to develop a relative and absolute localization schemes for the position estimate of stationary unattended ground sensors (UGS) node using a potential field method. The relative localization algorithm assumes that distance (i.e. range) measurements between sensor nodes are available and for absolute localization algorithm, some of the sensor nodes have absolute position information together with range measurement information. In the relative localization algorithm the UGS nodes are localized with respect to the co-ordinate system formed by the x-axis along the line connecting the first two sensor nodes. In absolute localization the UGS nodes are localized with respect to the known absolute position of some sensor nodes in the network. The localization algorithm is inspired from the potential field method used in mobile robots. We have implemented this algorithm on the Cricket mote network of our test-bed.

LabView toolkit for TinyOS
Despite their popularity, the Xbow motes still require a significant amount of time to be programmed, since a thorough knowledge of Tinyos software is required. Our Labview interface is able to install TinyOS programs (and also to allow the designer to change program parameters such as baud rate, mote ID and so on) on the Mote in a completely transparent way for the user. After the motes have been programmed, the read/write operations from/to the motes (either Cricket or MICA2) are completely performed in LabView. Also our interface provides an intuitive GUI for recording and visualizing the data in run-time and has an additional feature of performing data analysis such as FFT, Filtering, sound and vibration analysis etc.

Adaptive sampling
In this research project we use mobile robots as sensor carrying agents that facilitate the repositioning of network nodes to increase coverage, accuracy, and information gathering capabilities. The overall sampling objective is the accurate estimation of dynamically evolving environmental models subject to navigation, energy and communication constraints. Of particular interest will be the determination of optimal paths and mission-aware sampling strategies that effectively utilize available resources. Rather than sampling at pre-defined or random locations, the proposed Adaptive Sampling Algorithms (ASA) utilize information measures, estimation theory, and potential fields to direct the robots to the locations most likely to yield field variable information (Fig. 1). We combine classical robotic team concepts (goal attainment, flight formation, coverage) with traditional sensor network concepts (node energy minimization, optimal data rate, congestion control).

Congestion control
The congestion control problem in a network consists of finding (static optimization) and regulating (feedback stabilization) the optimal flow rates between nodes in the presence of network capacity constraints. Typically, solutions to both the routing and congestion control problems are differentiators between different network protocols, and by repositioning the mobile sensor nodes one can take full advantage of a given protocol. The approach proposed generates a potential field from the network utility function. Specifically, we would like to control individual robot location , and its sensor data rate over time in order to maximize a combined potential function. A relationship is established between each of the date sources and the individual links with data rates and capacity constraints through a routing matrix. In contrast to a typical static optimization in network flow control using constant capacity constraints, we allow the constraints to vary with the location of the robots, thus we maximize a utility function for all of the sources using fixed capacity constraints dependent on the location of the network nodes: subject to , where is a strictly concave utility function. Given a scalar potential field function U(r) that depends on the robot position, one can then calculate forces governing the robot motion of based on the gradient of the scalar potential field. After the force calculation, the equation of motion for the i-th sensor node will be given by integrating a mass-damper model over time . Finally, energy minimization considerations can be included using energy-dependent mass and damping coefficients.