Low-cost underwater robot development
The OpenROV is an inexpensive open underwater robotics experimental platform. Through evolutionary design modifications, we have been developing our own version of the popular platform. It currently features a stereo camera, IMU, compass, depth sensors, and ROS interfaces
Aerial radiation detection
A combined mathematical framework for visual inertial navigation, target tracking and interception, and active sensing tailored to nuclear measurement allows the detection of very weak radiation sources in transit order of magnitudes faster
Cooperative robot interactions
Robots do not have to avoid contact with other objects at all costs. Controlled physical interaction with their environment and other robots enables them to work with each other as a team to overcome limitations in their capabilities that do not allow them individually to complete their mission
Infant motor rehabilitation
Robots can become smart toys and interact with small children that have motor disabilities both physically and socially. Personalized models of child-robot interaction and advanced robot perception and control can boost rehabilitation dosage for these children
Planning and control under uncertainty
As the scale of robot design decreases, the level of noice and the intensity of environmental disturbancies does not follow at the same rate. The behavior of miniature robots can often be stochastic necessitating particular mathematical representations and control tools
Advanced mobility at the small scale
Hybrid designs give new capabilities to small-scale robotic platforms, extending their operational envelope
Locomotion modalities for rough terrain
New designs for robots that can negotiate challenging terrain with applications to agricultural and field automation
Robots on the farm
New designs for robust and multifunctional mobile robotic sensor platforms are deployed for automated maize (corn) phenotype measurements, especially under the canopy
Aerial systems autonomy
Aerial robotic systems are designed with completely autonomous flight capabilities: here a quadrotor UAV takes off and lands autonomously between two positions, sensing an obstacle while en route and reactively avoids it while flying at 2 meters per second. No map or GPS is used and perception is via an Intel Realsense camera.
Ground systems autonomy
Ground mobile robots navigate in outdoor environments without relying on external maps or GPS. Navigation and collistion avoidance is reactive based on sensor (LIDAR) input, and motion is controled through a receding horizon scheme that minimizes motion jerk. (The ethernet cable visible is for platform startup and data logging, in the absense of an outdoor adhoc wifi network.)
Autonomous Surface Vehicles
Autonomous marine robots have a wide range of applications, including environmental monitoring, infrastructure inspection and maintenance, research and rescue, and security. When designing controllers for them we may want to adjust our specifications so we optimize data collection. This Echoboat 160 carries a sonar for bathymetry mapping and is controlled so that its sensor experiences minimal accelerations to improve data quality.
Robots building robots
We explore the notion of machines building robots without human intervention, and the possibility that heterogeneous teams of these robots collaborate with each other by exploiting their diversity and combining their features in ways that brings out new capabilities that are not inherent in the constituent systems. Versions of the robots shown above can be produced with COTS components, fully inside a sophisticated 3D printer with pick-and-place capability, and move out of their build plate to autonomously interact with themselves and their environment.