Individual Differences and Bio-Inspired Design of Vehicle Group Dynamics
Abstract
Using an innovative and tightly integrated multi-disciplinary research approach, we willinvestigate the advantages and disadvantages to a wide range of tasks of individual differenceswithin groups. We will derive methodologies for distributed control of autonomous vehicles thatleverage, by design, the advantages in individual differences while avoiding the disadvantages.Our work will go well beyond the current literature on networked systems to address twoconnected goals: 1) to understand how individual differences among animals impact thecapabilities of animal groups in a range of scenarios and 2) to develop new bio-inspiredapproaches for autonomous vehicle groups that yield provable performance in complexenvironments.Naval missions require a group of vehicles to coordinate its activities, while successfullymanaging the significant challenges of a complex environment. This means a group must bestable in the presence of disturbance and flexible in its response to changes in the environment.It also must manage multiple tasks at the same time, for instance, source seeking and threatavoidance. These requirements create tensions: stability versus flexibility and source seekingversus threat avoidance. We will examine how these tradeoffs can be optimized by means ofdifferences among individuals in the group, for example, combining conservative individuals forstability with responsive individuals for flexibility.Our inspiration comes from quantitative studies of animal groups, where individualdifferences are increasingly being recognized as essential to effective functioning. A centralaspect of our approach will be to integrate the development of models and theory withexperimental work on real animal groups (schooling fish). We will introduce new ways tomodel, and experimentally quantify, differences in responsiveness, personality, state, sociability,and skill and incorporate multi-layer networks to represent multiple sensing modalities andinteractions associated to multiple tasks. We will examine dynamics that are fundamental tocritical Navy-relevant missions as well as collective animal behavior including: 1) spreadingdynamics, 2) decision-making and search dynamics, and 3) multiple task management.Experiments with fish schools will provide insights and inform model refinement. Tests in multivehiclelaboratories will aid development of control methodology for autonomous vehicle groupsthat emulate the remarkably resilient and versatile collective dynamics of animal groups.With our extensive and productive experience working closely together as a multidisciplinaryresearch team, we will leverage to great advantage the integration of our jointexpertise in biology, control theory, and robotics using experimentation, computation andanalysis.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 30, 2019
- Source ID
- N000141912556
Entities
People
- Naomi Ehrich Leonard
Organizations
- Office of Naval Research
- Trustees of Princeton University
- United States Navy