Enabling Large-Scale Collective Systems in Outdoor Aerial Robotics

For many real-life applications such as monitoring, mapping, search-and-rescue or ad-hoc communication networks, fleets of flying robots are expected to out-perform existing solutions. Robots can join forces to cover larger areas in less time, act as efficient communication relays and overcome difficult terrain more easily than ground-based robots. Compared to a single robot, the main advantages of collective systems are robustness, parallelism, flexibility, and the fact that they enable tasks that could not be solved by a single robot. While a multitude of theoretical models for collective operation have already been developed and tested in simulation, they have rarely been validated with physical robots yet. Transition to reality has so far been inhibited because of strong limitations in scalability. Indeed, the cost, safety risk, and number of operators associated with a collective aerial system increase proportionally with the number of robots. Substituting such systems for single robots is therefore unattractive or even impracticable. In previous experimental approaches, a maximum of five physical robots were operated simultaneously in outdoor scenarios, assisted by human backup pilots on the ground. In contrast, most theoretical models rely on a minimum of ten robots to obtain interesting swarm effects. With respect to this discrepancy, we consider here ten robots to be a large-scale system in aerial robotics. In order to scale real collective aerial systems to at least ten robots, we suggest the following paradigm: flying robots must be low-cost, inherently safe, deal with mid-air collisions and require minimal supervision from an operator on the ground, such that the entire system warrants similar cost as well as similarly safe and easy deployment as a single, classical flying robot. Moreover, this would make swarms of flying robots as accessible as wheeled robots that have already been used successfully in larger collective systems. The above criteria can best be addressed with a global, systematic approach on all major design levels, which are the robot's airframe, low-level control, collective supervision and control as well as mid-air collision avoidance. On each level, we identified the bottlenecks related to scaling and developed methods for compliant robot design. Based on a systematic analysis of airframe configurations, a flying-wing is proposed that is made of low-cost, safe and durable foam material and can be deployed everywhere by hand-launch. A model for the dimensioning of such an airframe is presented. For low-level control, a novel minimalist control strategy is proposed, implemented on an inexpensive, custom-made autopilot and validated in field tests. In order to enable robots for collective operation, we suggest to use WiFi communication links and embedded Linux-computers onto which swarm designers can download distributed controllers for collective behaviors directly from their computer simulation. Furthermore, the idea is put forward to enable collective supervision through an operator interface with the modality for direct group-control. Implemented and field-proven, these solutions can serve as guidelines for other developers. Finally, the problem that robots may collide with each other during collective operation has been addressed with a model assessing the collision risk and a distributed strategy for mid-air collision avoidance, validated on real robots. Applying the proposed methodology allowed us to demonstrate the world's first collective system composed of ten autonomous flying robots in outdoor experiments with several different collective behaviors. Robots are low-cost (about 1/10 the cost of commercially available robots), inherently safe (kinetic energy of a medium-sized bird) and can be configured, deployed and supervised by a single operator on the ground. This represents a significant step for the scaling of flying collective systems with respect to the state of the art as well as an unprecedented operator-to-robot ratio.

Communication in a Swarm of Miniature Robots: The e-Puck as an Educational Tool for Swarm Robotics

Christopher Cianci, Alcherio Martinoli, James Pugh, Xavier Raemy

Swarm intelligence, and swarm robotics in particular, are reaching a point where leveraging the potential of communication within an artificial systempromises to uncover newand varied directions for interesting research without compromising the key properties of swarmintelligent systems such as self-organization, scalability, and robustness. However, the physical constraints of using radios in a robotic swarm are hardly obvious, and the intuitive models often used for describing such systems do not always capture them with adequate accuracy. In order to demonstrate this effectively in the classroom, certain tools can be used, including simulation and real robots. Most instructors currently focus on simulation, as it requires significantly less investment of time, money, and maintenance—but to really understand the differences between simulation and reality, it is also necessary to work with the real platforms from time to time. To our knowledge, our coursemay be the only one in the world where individual students are consistently afforded the opportunity to work with a networked multi-robot system on a tabletop. The e-Puck,1 a low-cost small-scale mobile robotic platform designed for educational use, allows us bringing real robotic hardware into the classroom in numbers sufficient to demonstrate and teach swarm-robotic concepts.We present here a custom module for local radio communication as a stackable extension board for the e-Puck, enabling information exchange between robots and also with any other IEEE 802.15.4-compatible devices. Transmission power can be modified in software to yield effective communication ranges as small as fifteen centimeters. This intentionally small range allows us to demonstrate interesting collective behavior based on local information and control in a limited amount of physical space, where ordinary radios would typically result in a completely connected network. Here we show the use of this module facilitating a collective decision among a group of 10 robots.

2007

Synthesis, modeling, and experimental validation of distributed robotic search

James Pugh

The work of this thesis centers around the research subject of distributed robotic search. Within the field of distributed robotic systems, a task of particular interest is attempting to locate one or more targets in a possibly unknown environment. While numerous studies have proposed and analyzed methods to accomplish this, there is currently a lack of a strong foundation of tools and techniques that can be used to facilitate the development and evaluation of different approaches to distributed robotic search. In this work, we aim to provide such a foundation through tools, methods, models, and analysis of experimentation. An often overlooked aspect of the distributed robotic research process is the development and analysis of tools and modules to be used with robotic systems. These may include plug-ins for realistic robotic simulators, software/hardware systems to track multiple mobile robots in real-time, extension boards for robotic platforms, and the robots themselves. Along with other tools developed and used for this work, we focus particularly on the development, characterization, and validation of a fast, accurate on-board system for relative positioning and communication between robots, a capability which is critical for effective distributed robotic search. Designing individual robot controllers to generate a specific group behavior is a difficult and often counter-intuitive process. A possible alternative to hand-crafting distributed search controllers is automatic synthesis using machine-learning techniques. We explore the effectiveness of using a noise-resistant version of the Particle Swarm Optimization algorithm to optimize the weights of an embedded Artificial Neural Network, allowing the robot to learn obstacle avoidance behavior (a common benchmark for robotic learning techniques); we find that this technique appears to offer superior performance as compared to the canonical approach of using Genetic Algorithms for this type of learning. A method for faster learning using distributed evaluation in a robot team is tested and is found to offer comparable performance using only a fraction of the original learning time. This technique can be used for fast, effective learning and adaptation in a distributed robotic system performing search. The process of designing and analyzing algorithms for distributed robotic systems can be greatly facilitated if models are available to describe the dynamics of the algorithm at an abstract level. Inspired by previous examples in the distributed robotic field, we work to design a model of robotic search that captures the system at different levels of abstraction, ranging from accurately recreating the details of individual robots to describing the entire system as an indivisible whole. To capture the entire search process, we model both the exploration phase, where robots cover an environment in an effort to detect traces of targets, and the localization phase, where robots use target emission sensing to navigate towards the target. The utility of our models is demonstrated by using them to develop an effective technique for the declaration phase of search, where robots decide that a target has been accurately localized and announce its position. In distributed robotics research, it is important that techniques developed with abstracted simulations and models are ultimately validated using real robotic platforms in order to verify their correctness. In that spirit, we run systematic sound search experiments using teams of up to ten real robots. These experiments utilize the tools developed throughout the research process, demonstrate the utility of our learning technique for fast search adaptation, and serve to validate our models of distributed robotic localization. In addition, they allow us to analyze and better understand the subtle dynamics of the search process, providing information which should be useful for any future work on distributed robotic search.

EPFL2008

Enabling Collective Operation of Indoor Flying Robots

James Roberts

Collective flying robots show great potential in many diverse indoor applications. Their added robustness, parallel operation and redundancy has clear advantages over a single flying robot. However, flying within cluttered and unprepared indoor environments, even for a single flying robot, is extremely challenging. The main reason why collective flying robots have not yet been successful within indoor environments, is due to a combination of several challenges related to the size constraints placed on an indoor hovering platform, which directly limits the available energy, embedded sensing and processing capabilities of a flying robot. The energetic cost of flying, places limits on the flight endurance and practicality of a swarm of flying robots. The current spatial-coordination approaches implement methods that are either too computationally expensive, or impractical for real-world operation, within unknown and unprepared indoor environments. The goal of this thesis was to develop a practical methodology for enabling energy efficient, autonomous indoor flying robots capable of inter-robot spatial-coordination for unprepared indoor environments, without using external aids. In order to enable collective operation of indoor flying robots, several practical methodologies have been proposed and demonstrated. A generalised design strategy has been proposed to dimension a hovering platform for a specific flight endurance, payload capability and robustness criteria. The developed method can be used as a practical design tool for anyone working with hovering platforms. The dimensioning strategy has created a design, which is highly suitable for carrying the necessary sensing and processing required to enable the collective operation of indoor flying robots. A simple sensing and control strategy is proposed for enabling anti-drift control and obstacle avoidance behaviours on an indoor highly dynamic, hovering platform. The approach has enabled one of the first indoor hovering platforms that could achieve such a capability without using any external aids. For the collective operation of indoor flying robots to work in reality, the on-board energy needs to be managed efficiently and conserved in a way that allows a swarm of robots to be useful, extending beyond the individual 10-20 min flight time. A generalised energy model has been developed, allowing for the accurate estimation of the flight endurance and perching time of hovering platforms. The energy model can be used to optimise the battery selection process of a hovering platform, to obtain the highest possible endurance. This is the only model known that is able to predict any combination of flying endurance and perching times. Additionally, a method of attaching to the ceiling has been presented that allows a flying robot to conserve energy and have a stable birds-eye-view while performing static sensing tasks. By applying energy management techniques, through use of energy modelling and behaviours that reduce the flight time, the energetic cost of flying can be mitigated and the mission endurance can be extended over several hours, which is especially useful for collective operation. A new infrared ranging technique has been developed that allows for a high sensing performance, including long range (12 m), high-speed (1 kHz / # robots) and high resolution (better than 1.1 cm up to 6 m). A practical on-board sensing method using this technique, has been developed that can provide spatial-coordination between multiple robots in three dimensions. The developed approach allows for easily adaptation, to suit other robots and applications, depending on a specific sensing speed and coverage requirement. The developed sensor is the worlds first embedded 3-D relative positioning sensor that has the ability to enable inter-robot spatial-coordination in three dimensions, which is necessary for achieving goal-directed flight on highly dynamic flying robots. A practical autonomous flight control methodology has been demonstrated that can provide hovering platform stabilisation, 3-D obstacle avoidance and 3-D waypoint navigation, all using the 3-D relative positioning sensor. Goal-directed flight and collective deployment have been achieved using only the information from the 3-D relative positioning sensing. The developed methodologies within this thesis, has enabled for the first time, the collective operation of highly dynamic indoor flying robots, without using external aids.

EPFL2011