ACS_facuty SEMINAR ANNOUNCEMENT: Swaroop Darbha "Persistent Monitoring with the aid of UAVs" Friday, November 15, 2019 at 2pm in POTR, room 234.

[Dave Cappelleri]

ACS Affiliated Faculty-


This seminar may be of interest.




%%%%%%%



*Persistent Monitoring with the aid of UAVs *




*Swaroop Darbha *Professor of Mechanical Engineering

Texas A&M University, College Station, TX-77843

Friday, November 15, 2019

2pm in Potter Fu Room, 234



*Abstract*



Many civilian and military applications require close monitoring of
specified locations (referred to as targets) with the aid of a UAV.
Typically, a UAV takes off from a depot, visits/revisits the set of targets
in a specified sequence and lands at the depot, gets serviced
(refuelled/recharged) and continues the same “walk” again and again for
persistent monitoring. In some communication-constrained applications, the
data collected by UAVs must be physically transported to the depot; data
can also be transmitted by the UAV only in the proximity of the depot while
in flight. Freshness of data required for follow-on actions demand that
depot, like the targets, must be visited frequently and hence, the depot
will also be treated as a target. Performance of monitoring is
characterized by (1) maximum elapsed time between visits to the same
target, and (2) the mission completion time.  For circumventing the
complexities of modeling fuel/battery consumption,  the on-board power
limitation is surrogated by the maximum number V of visits in a walk. In
this talk, the following problem will be considered: Find an optimal walk
of k <= V visits so that a weighted combination of walk revisit time
(maximum time elapsed between any successive revisits of any target)  and a
penalty psi(V-k) is minimized. The penalty term psi(V-k) discourages a UAV
to get serviced earlier than V visits as servicing increases  the total
mission time by (V-k) times the servicing time for every kV walks.
Efficient algorithms will be presented that exploit the structure of the
optimal solution when the servicing time is negligible; this case also
provides useful lower bounds for the given problem and forms the basis for
an analytical result and efficient algorithm when the servicing time
exceeds a threshold.



*Bio*



Swaroop Darbha received his B. Tech from IIT-Madras in 1989 and PhD from UC
Berkeley in 1994; his doctoral work deals with decentralized control for
string stability of vehicle platoons. He was a post-doctoral researcher
with the California PATH program till 1996 working on various aspects of
Automated Highway System design. He joined the faculty of mechanical
engineering at Texas A&M University in 1997, where he is currently a
professor. His research has focused on ground Autonomous and Connected
Vehicles (ACVs) and UAVs. In ACVs, his research spans vehicle modeling,
control and diagnostics for automatic vehicle following systems, traffic
flow modeling with ACVs, and assessment of traffic mobility and safety
benefits due to connectivity with ACVs. His collaborative work on UAVs with
AFRL spans two decades and has worked on combinatorial motion planning of
UAVs with motion, mission, communication and resource constraints,
decision-making under uncertainty for UAVs in the context of perimeter
surveillance, ISR and intrusion detection. He is a fellow of ASME and IEEE
for his contributions to Intelligent Transportation Systems (ITS) and
Unmanned Vehicles.




*Host Professor Shreyas Sundaram, sundara2 at purdue.edu
<sundara2 at purdue.edu>, 49-60406*



-- 


David J. Cappelleri, Ph.D.
Associate Professor
Purdue University
Mechanical Engineering
585 Purdue Mall
West Lafayette, IN  47907-2088
E-mail: dcappell at purdue.edu
Phone:  765.494.3719
Fax:    765.494.0539
website:  www.multiscalerobotics.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: </ECN/mailman/archives/acs_faculty/attachments/20191112/5d0a261f/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Seminar- Swaroop Darbha.pdf
Type: application/pdf
Size: 153061 bytes
Desc: not available
URL: </ECN/mailman/archives/acs_faculty/attachments/20191112/5d0a261f/attachment-0001.pdf>