[robotics-worldwide] 2011 Marine Robotics/AI Planning & Execution, Machine Learning Summer Internship, Monterey Bay California

Kanna Rajan kanna.rajan at mbari.org
Wed Dec 1 11:48:51 PST 2010

[Apologies if you receive multiple copies of this announcement]

MBARI's Autonomous Systems group has an opening for one internship  
position for the summer of 2011 for a beginning PhD student.


MBARI (Monterey Bay Aquarium Research Institute http://www.mbari.org)  
is a private non-profit oceanographic research institute which has  
engaged in inter-disciplinary research in the ocean sciences and  
technology for over 20 years. The institute does advanced work in  
Robotics, Genetics, Micro-Biology and Ocean Exploration and Observing  
with 3 ocean going ships, 2 Remotely Operated Vehicles (ROV's) and 3  
Autonomous Underwater Vehicles (AUV's) in addition to state-of-the-art  
science and technology facilities.  2011 Internship applications are  
now open with details at:


These internships are competitive, paid and open to all nationalities.  
They are for a fixed period of 10 weeks starting June 13th 2011.

MBARI's Autonomous Systems group is engaged in research efforts to use  
advanced Artificial Intelligence (AI) techniques to enable Autonomous  
Underwater Vehicles (AUV http://www.mbari.org/auv/) to be more  
adaptive and self aware of their resource and goal commitments. Our  
onboard deliberative system T-REX (http://www.mbari.org/autonomy/ 
TREX/) being used onboard an AUV is based on the sense-deliberate-act  
paradigm synthesizing new mission plans in response to environmental  
conditions. T-REX uses a state-of-the-art temporal constraint-based  
planner originally developed at NASA and flown on a number of space  
missions including the ongoing Mars Exploration Rovers (MER) mission.  
All projects are associated with the CANON (http://www.mbari.org/ 
canon/) program and are research centric.


Potential internship opportunities for 2011 include working on  
Constraint-based Planning, Machine Learning and Spatial Statistics.  
Potential projects are along the lines of:

1. Automated Planning: Dealing with discrete resources for planning  
within the constraint-based planning framework reflecting the  
consumables onboard an AUV. Modeling and integrating the use of these  
resources within the plans synthesized by TREX is an important problem  
for these resource restricted platforms which work in harsh  

2. Automated Planning: Using constraint-based temporal planning  
techniques using coarse grained plan representation to coordinate  
multiple assets (robots, ships, aircraft fly-bys) for oceanographic  
field experiments.

3. Machine Learning: Synthesis of multiple Hidden Markov Models (HMMs)  
for discovering unstructured and dynamic features in the coastal ocean  
is critical to make optimal use of the platform at sea. Currently T- 
REX uses only a single model; interrupting plan execution to chase  
multiple hypothesis at sea by an AUV is the core of the effort.

4. Machine Learning: Techniques in parameterizing incoming data  
streams from heterogenous sources for event response. Such data will  
be used for tagging by mixed human and automated mechanisms for  
detection triggers of dynamic, episodic and non-deterministic features  
in the coastal ocean. Techniques popularized by Amazon's Mechanical  
Turk in the general area of Folksonomy will be used for identifying  
such features for eventual investigation by autonomous robots.

5. Spatial Statistics: Model building of dynamic ocean features which  
incorporate space and time and doing so onboard a resource limited  
robot with implications for adaptive sampling.


Highly motivated beginning PhD students with substantial

+ Java, C++ and MATLAB programming skills; this is a must have
+ AI Planning and a background in Constraint Satisfaction
+ Statistical Machine Learning backgrounds
+ Self motivated with good communication skills

and desirous of making a research impact in the ocean sciences are  
encouraged to apply. The interns will be expected to go to sea at  
least once during the fellowship. Applicants must be registered in a  
PhD program. 3 letters of support are necessary. Only applicants with  
a focus on AI and Robotics related projects will be considered.

Previous visiting interns have come from MIT (4), Colorado State,  
University of Michigan, Strathclyde University (UK), University of   
Birmingham (UK), LAAS/Toulouse (France), Univ. of California Santa  
Cruz, Univ. of Southern California (2), Univ. Carlos III de Madrid  
(Spain) (2), IIT Madras (India) and Heriot-Watt University (UK).

Women and Minorities are particularly encouraged to apply.

More information about the robotics-worldwide mailing list