Co-Optimization of Sensing, Communications and Navigation

Co-Optimization of Sensing, Communications and Navigation of a Robotic Network under Resource Constraints

  • Project Duration: 2013-2017

  • Current Team Members:

  • Supporting Grant:

    This project is supported by NSF NeTS award # 1321171.

Related Publications

Back to top

Our Approach

The main goal of this research is to develop a foundation for the co-optimization of sensing, communication and navigation in robotic networks. More specifically, we want to utilize the fact that the robot has control over its mobility to actively enforce the needed connectivity via communication-aware navigation designs. Furthermore, we are interested in an energy-aware operation. Thus, our second major objective is to show when the robot should incur motion energy to achieve better connectivity and when it should incur communication energy (by increasing its transmit power). In our third major goal, we are interested in developing computationally-efficient approaches for path planning, by utilizing space-filling curve, in order to address the underlying traveling salesman problem that occur in several of the robotic network problems.

So far, we have proposed a new way of decision making which allows each robot to co-optimize its sensing, communication and navigation objectives. This is truly a multi-disciplinary approach that combines communication theory with robotics. Our approach relies on developing probabilistic predictors for connectivity at unvisited locations, which then enables the robot to find the trajectory that gathers the most information out of the environment (via sensing) while maintaining the needed connectivity. We have furthermore proposed a framework to enable a team of unmanned vehicles to provide connectivity in harsh environments through cooperative communication such as robotic routing and robotic beam forming. Then, the robots can properly control their motions to move to spots better for cooperative communication using our probabilistic link metrics.

We have shown how to extend the current literature on distributed decision making via binary log-linear learning (which mainly focuses on ideal communication links) to consider the impact of stochastic communication channels. More specifically, we have driven conditions on the probability of link connectivity to achieve a target probability for the set of potential maximizers of the game. This will ensure that the global task will be accomplished via local decision making despite the stochasticity of the links.

We have furthermore considered the energy consumption of the network. In our "To Go or Not to Go" problem, we have shown when the robot should incur motion energy to move to a spot better for connectivity. In contrary to the general belief that motion energy is always much more expensive and thus the robot should always increase communication power to achieve better connectivity, we have shown that there are several cases where it is more beneficial for the robot to incur motion energy and move to a spot better for connectivity. We have validated our theoretical findings with real motion parameters of pioneer robots as well as real channel measurements. Our results so far indicated that we can cut the energy consumption in half.

Any robotic task that involves visiting a number of sites for information gathering will have a combinatorial nature, similar to a traveling salesman problem. In addition, the robotic network has several other constraints such as a given energy budget, navigational constraints, connectivity requirements, and operation time budget. Thus, the computational complexity of finding the optimal solution can be prohibitive. Along this line, we have proposed a new efficient approach, based on using space-filling curves. Our approach can achieve a near-optimum solution with a considerable reduction in the complexity.

Back to top


The overall outreach goal of this proposal is to expose Native American and Hispanic students to the basics of wireless systems and robotics. In the summer of 2015, PI's lab hosted a Veteran Native American community college student in her lab and trained a female graduate student to be his mentor. The outcome was very fruitful and fulfilling to both the mentor and the student. See below for sample pictures. In 2014, we have also organized a very successful outreach event this year. More specifically, in the spring of 2014, we arranged for 5 students and 2 teachers of the Gallup Central High School to fly to UCSB for a workshop for minority students. Gallup Central High School is at the border of New Mexico and Arizona and mainly consists of Native American and Hispanic students. Through our several discussions, it became clear that an educational/research camp at UCSB could have a tremendous impact on the lives of these students. Thus, we worked with the SACNAS program (minority program) at UCSB to have a few students and teachers come from Gallup Central High School as part of UCSB SACNAS annual event.

The event became an instant success and very rewarding. 5 students and 2 science teachers attended a 3 day workshop, got to interact with our lab, see our experimental robotic setup, attended several lectures by different faculty, toured the campus and got inspired by college life. Gallup Central High School was very thankful for the experience, which turned out to be one of the most rewarding outreach activities that our lab has done.

As part of this event, the PI further gave two lectures on robotics to not only these students, but also to all the incoming minority students from several different local schools (around 50 students). Our lab also held two robotic demo sessions for all the incoming SACNAS students.

The pictures below highlight some of the moments of our outreach events.

Back to top