Oscar De Silva

Development of a relative localization scheme for ground-aerial multi-robot systems

In this paper we demonstrate a design and experimentation of a relative localization solution for a multi robot team involving both ground and aerial robots. The relative localization method proposed in this paper has the ability to localize a dynamic agent with respect to only one leader ground robot in a GPS denied environment. The sensor solution proposed in the study employs a combination of an acoustic sensor and an infra-red(IR) based vision sensor for relative range and bearing estimations respectively. An extended Kalman filter performs the sensor fusion using a four degree of freedom kinematic model. Numerical simulations validate the sensor fusion scheme for both ground and aerial robotic relative localization. An experimental test-bed of the system with the hardware implementation of the sensors were developed. For comparison purposes the self localization modules of the robots are further integrated into the experimental setup. Realtime experiments were performed where 5-10 cm mean accuracy of pose estimation was achieved in multiple experiments.

An Ultrasonic and Vision-Based Relative Positioning Sensor for Multirobot Localization

This paper proposes a novel 3D sensor node to establish relative measurements within a robot network. The developed sensor nodes employ ultrasonic-based range measurement and infrared-based bearing measurement for spatial localization of robots. The sensor is low power, lightweight, low cost, and designed to be applicable across many robotic platforms, including microaerial vehicles. The proposed sensor design requires only two robots to perform relative measurements of each other and achieves a measurement accuracy of 0.96-cm Root-Mean-Square Error (RMSE) for range and 0.3° RMSE for bearing. The sensor nodes are scalable and can be configured using either Star or Mesh protocols with a maximum of 10-Hz update rates over a detection range of 9 m. The correspondence issue of having multiple robots is resolved using time division multiple access methods where different time slots are used by each sensor node. These features are verified by multiple experimental evaluations on a multirobot team with both ground and aerial agents. The proposed approach allows multirobot localization in scenarios where supportive positioning services such as GPS are unavailable. As a result, even basic robots, which lack powerful simultaneous localization and mapping capabilities, will be capable of autonomous navigation by accessing the positional information provided by the sensor network.

Pairwise observable relative localization in ground aerial multi-robot networks

This paper addresses the problem of relative localization in a team of robots which consists both ground and aerial platforms. The robots are equipped with sensors for measuring both range and bearing of neighboring team members. Pairwise observability in such a team refers to the ability of two robots to estimate their relative poses, without strictly relying on information or measurements of other team members. This capability is important to realize many robotic behaviors such as sense and avoidance, formation control, and leader follower supervisory control, in a robust and minimally dependent manner. This paper presents an implementation and an analysis of a multi-robot relative localization network. In order to identify the necessary conditions for achieving pairwise observability, the study performs a nonlinear observability analysis. The analysis considers the cases where input velocities of the measured platforms are unknown, which is relevant to most drifting aerial platforms facing communication constraints and sensing limitations. The results of the analysis are experimentally demonstrated, along with the implications of the observability study in designing multi-robot teams and estimation frameworks.

Efficient distributed multi-robot localization: A target tracking inspired design

The main reported solutions for the problem of multi-robot relative localization require synchronous communication between robots, where the network should communicate each time a relative measurement is logged in the team. This paper proposes a localization method, which can accommodate communication at a low predefined rate rather than forcing communication each time a measurement is logged. This is achieved without explicitly accumulating past measurements locally at each robot. This capability is necessary to support increasing number of robots in a team, under finite communication and computation resources. The design includes a novel fusion strategy, a consistent estimation method, and a state based initialization method, embedded in a distributed target tracking framework. The design is efficient in terms of computation demand, since it scales linearly with the number of robots. Additionally, the design is efficient in terms of communication demand, since communication is neither required to be synchronized with sensor readings, nor constrained to a specific network topology. The paper validates the proposed approach for its initialization capability, consistency of estimates, and robustness of performance, through several numerical simulations and using a publicly available multi-robot data set.

Relative localization with symmetry preserving observers

Symmetry preserving observer design is a recently developed approach for deriving estimators which exploits the invariant properties of nonlinear systems. Multi-robot localization is inherently a system operating on SE(3), thereby posing an interesting problem for application of the filter. This paper presents an invariant extended Kalman filter design for the problem of multi-robot relative localization in 2.5D, for application in ground and aerial mobile platforms. A detailed derivation of the invariant filter is presented with numerical results analyzing its performance against a traditional EKF approach to the problem. The tracking errors, stability to random initialization and robustness to changing noise characteristics are evaluated. The strong geometric basis of the filter results in linear Kalman like gain convergence behavior which is desirable for numerical stability and applicability as a low cost scheduled gain observer to the problem.

The Right Invariant Nonlinear Complementary Filter for Low Cost Attitude and Heading Estimation of Platforms

This paper presents a novel filter with low computational demand to address the problem of orientation estimation of a robotic platform. This is conventionally addressed by extended Kalman filtering of measurements from a sensor suit which mainly includes accelerometers, gyroscopes, and a digital compass. Low cost robotic platforms demand simpler and computationally more efficient methods to address this filtering problem. Hence nonlinear observers with constant gains have emerged to assume this role. The nonlinear complementary filter is a popular choice in this domain which does not require covariance matrix propagation and associated computational overhead in its filtering algorithm. However, the gain tuning procedure of the complementary filter is not optimal, where it is often hand picked by trial and error. This process is counter intuitive to system noise based tuning capability offered by a stochastic filter like the Kalman filter. This paper proposes the right invariant formulation of the complementary filter, which preserves Kalman like system noise based gain tuning capability for the filter. The resulting filter exhibits efficient operation in elementary embedded hardware, intuitive system noise based gain tuning capability and accurate attitude estimation. The performance of the filter is validated using numerical simulations and by experimentally implementing the filter on an ARDrone 2.0 micro aerial vehicle platform.