Aerial Swarm Robotics Introduction and Exploration Parth Sarthi - PowerPoint PPT Presentation

MIN-Fakultät Fachbereich Informatik Aerial Swarm Robotics Introduction and Exploration Parth Sarthi Pandey Universität Hamburg Fakultät für Mathematik, Informatik und Naturwissenschaften Fachbereich Informatik Technische Aspekte MultimodalerSysteme 11. December 2017 P.S. Pandey – Exploring Aerial Swarms

Outline 1. Motivation From fellow Earthlings 2. What are swarms? Definition & Properties 3. Surveillance by Micro-Aerial Vehicles Motion Planning Particle Swarm Optimization 4. Energy Aware PSO Algorithm 5. Aerial Swarms as Asymmetric Threats 6. Conclusion P.S. Pandey – Exploring Aerial Swarms 2/ 27

Motivation Sapiens: The reason why Homo Sapiens are the only human species alive today and ruling the planet is their capacity to coordinate in large numbers and work together for a commongoal . P.S. Pandey – Exploring Aerial Swarms 3/ 27

Motivation [www.fuelspace.org] P.S. Pandey – Exploring Aerial Swarms 4/ 27

Motivation [iopscience.iop.org] P.S. Pandey – Exploring Aerial Swarms 5/ 27

Motivation [www.treehugger.com] P.S. Pandey – Exploring Aerial Swarms 6/ 27

Motivation Beehive Wasp Nest [agriculture.purdue.edu] [en.wikipedia.org] P.S. Pandey – Exploring Aerial Swarms 7/ 27

How do we define a swarm?  Not just a random group of agents.  Properties:  No centralized controlling  Only local sensing and communication  Large number of agents  Single agent relatively incapable [spectrum.ieee.org/automaton/robo tics/robotics-hardware/a-thousand- kilobots-self-assemble] P.S. Pandey – Exploring Aerial Swarms 8/ 27

Collective Behavior Collective behavior is classified into 4 categories:  Coordination: appropriate organization in space and time  Cooperation: individuals achieve tasks together which could not be done by a single one alone  Deliberation: colony faces several options and collectively choses one of them  Collaboration: different activities simultaneously performed by groups of specialized individuals Garnier, Simon, Jacques Gautrais, and Guy Theraulaz. "The biological principles of swarm intelligence." Swarm Intelligence 1.1 (2007): 3-31. P.S. Pandey – Exploring Aerial Swarms 9/ 27

Aerial Swarms  Multiple unmanned aerial vehicles.  Coordinating the space and time  Working together for a single task  Collectively taking real time decisions  Collaborating with one another A common COTS quadcopter widely used in research and by hobbyists. [Wilkerson et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 10/ 27

Surveillance by Micro-Aerial Vehicles Applications of Aerial Swarms Initial position of MAV. Obstacles – Green lines. No fly zones – red lines. Areas of interest – Blue regions. [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 11/27

Surveillance by Micro-Aerial Vehicles MAV Group deployed. Camera range – white pyramids. Trajectories – Yellow curves. Localization linkages – Red lines. [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 12/27

Swarm Motion Planning  Two main challenges:  How to find a suitable swam distribution that covers all AoIs.  How to find feasible trajectories of MAVs – described by Swarm Distribution.  Cost Function:  Describes the quality of coverage of AoIs achieved by a swarm distribution X.  Is a vector of 3D positions of ‘n’ MAVs [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 13/ 27

Construction of a Map [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 14/27

Construction of a Map [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 15/27

Optimizing Swarm Distribution  Challenges:  Each MAV represented by its 3D position.  ‘n’ MAVs belong to a swarm.  Cost Function may have several local minima .  High dimensional optimization problems  Plenty of local extremes.  Particle Swarm Optimization [Saska et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 16/27

Particle Swarm Optimization  Eg. Swarm of birds searching for food.  Swarm as a vector of P particles.  Each particle keeps track of its “best” position  “pbest” for individual particle.  “gbest” for best in population.  “lbest” for best in defined neighborhood.  At each time step, each particle stochastically accelerates toward its pbest and gbest or lbest. [Kennedy & Eberhart 2001] P.S. Pandey – Exploring Aerial Swarms 17/27

Particle Swarm Optimization Process 1. Initialize population in hyperspace. 2. Evaluate fitness of individual particles. 3. Modify velocities based on previous best and global (or neighborhood) best. 4. Terminate on some condition. 5. Go to step 2. [Kennedy & Eberhart 2001] P.S. Pandey – Exploring Aerial Swarms 18/27

Energy Aware Particle Swarm Optimization Towards Efficiency EAPSO – considers the trade-off between profit and energy consumption. [Mostaghim et. al. 2016] Mostaghim, Sanaz, Christoph Steup, and Fabian Witt. "Energy Aware Particle Swarm Optimization as search mechanism for aerial micro-robots." Computational Intelligence (SSCI), 2016 IEEE Symposium Series on. IEEE, 2016. P.S. Pandey – Exploring Aerial Swarms 19/27

Building Blocks of EAPSO  DecideState – Takeoff or Stay Grounded  LeaderSelection – Choses Best Individual (PSO or Local Search)  ComputeVelocity , UpdatePosition , ComputeEnergy [Mostaghim et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 20/27

It's not just Rainbows and Butterflies Aerial Swarms as Asymmetric Threats Estimated sales of popular drones Intel Drone 100 record breaking aerial swarm. (i.e. quadcopters) [Wilkerson et. al. 2016] Wilkerson, Stephen, et al. "Aerial swarms as asymmetric threats." Unmanned Aircraft Systems (ICUAS), 2016 International Conference on. IEEE, 2016. P.S. Pandey – Exploring Aerial Swarms 21/27

Aerial Swarms as Asymmetric Threats  Cheaper and more powerful drones readily available.  Intel set the Guinness World Record for most simultaneously airborne UAVs.  US Navy program LOCUST (Low-cost UAV Swarming Technology)  Plans to include armed and unarmed UAVs  Aerial Swarm Attack a much bigger threat than acknowledged for. [Wilkerson et. al. 2016] P.S. Pandey – Exploring Aerial Swarms 22/27

Conclusion  Motivation from nature – Awesome  Application in UAV Surveillance – Awesomer  Energy Efficient Enhancements to the Tech – Awesomerer  But…  Threats are Real – Aerial Swarm Attack (Boomm!!^n)  Needed..  Investments in countermeasures.  Containing the virus before it contaminates. P.S. Pandey – Exploring Aerial Swarms 23/27

References [Garnier et. al. 2007] Garnier, Simon, Jacques Gautrais, and Guy Theraulaz. "The biological principles of swarm intelligence." Swarm Intelligence 1.1 (2007): 3-31. [Saska et. al. 2016] Saska, Martin, et al. "Swarm distribution and deployment for cooperative surveillance by micro-aerial vehicles." Journal of Intelligent & Robotic Systems 84.1-4 (2016): 469-492. [Mostaghim et. al. 2016] Mostaghim, Sanaz, Christoph Steup, and Fabian Witt. "Energy Aware Particle Swarm Optimization as search mechanism for aerial micro-robots." Computational Intelligence (SSCI), 2016 IEEE Symposium Series on. IEEE, 2016. [Wilkerson et. al. 2016] Wilkerson, Stephen, et al. "Aerial swarms as asymmetric threats." Unmanned Aircraft Systems (ICUAS), 2016 International Conference on. IEEE, 2016. P.S. Pandey – Exploring Aerial Swarms 26/27

References (contd..) uhhBeamer [Kennedy & Eberhart 2001] Eberhart, Russell C., Yuhui Shi, and James Kennedy. Swarm intelligence. Elsevier, 2001. [i] www.fuelspace.org [ii] iopscience.iop.org [iii] www.treehugger.com [iv] agriculture.purdue.edu [v] en.wikipedia.org [vi] https://spectrum.ieee.org/automaton/robotics/robotics- hardware/a-thousand-kilobots-self-assemble [vii] dailymail.co.uk P.S. Pandey – Exploring Aerial Swarms 27/27