LILI: Lehigh Instrument for Learning Interaction
LILI is a joint project with the Lehigh WiNS Lab lead by Prof. Mooi Choo Chuah. Our goal is to develop a low-cost, interactive robot that can be used in a home environment to work with children who have autism spectrum disorder. LILI interacts with users via gestures, voice commands, and an animated speaking avatar. LILI can recognize users' faces, and her motion can be controlled either via gesture or voice.
RoSCAR: Robot Stock Car Autonomous Racing
RoSCAR is a new, low-cost, high-performance mobile robot platform designed for use in education, research, and competitions. The RoSCAR platform is based on a 1/10 scale short track race car, integrated with an on-bard desktop-class computer, odometry, and RGB-D sensing.
Dynamic soaring is a technique whereby horizontal wind that varies in strength or direction is used to support flight. Seabirds like albatrosses are known to travel hundreds of kilometers in a single day utilizing dynamic soaring. We are investigating the possibility of using dynamic soaring techniques and solar power to generate perpetual flight of a Unmanned Aerial Vehicle (UAV) in the jet stream. This is a joint project between Lehigh University and Penn State University, and draws upon expertise in robotics, mechanical engineering, and aerospace engineering.
The Smart Wheelchair Project
Building on the lessons learned in creating the Automated Transport and Retrieval System(ATRS) and the DARPA Urban Challenge, this project aims to move the wheelchair away from the vehicle and give the wheelchair its own level of autonomy. By leveraging the high fidelity sensors on Little Ben (Ben Franklin Racing Team's entry to the DARPA Urban Challenge) we plan to create high resolution 3D landmark maps. The wheelchair then can navigate autonomously inside these maps with a less expensive sensor suite. This project focuses on creating the algortihms, techniques, and software for accomplishing this goal while using readily available sensors and equipment.
The Lehigh Mapping Trike
The Lehigh Mapping Trike (LMT) project was motivated by the lack of means to construct large-scale, three-dimensional maps in outdoor pedestrian zones as required for The Smart Wheelchair Project. The LMT features 3 Sick LMS219 LIDARs, GPS, encoders, an inertial measurement unit, and onboard computing. Sensor data are fused using SLAM algorithms to maintain an accurate localization estimate, and scans from the vertically scanning LIDARs are registered to the vehicle's position over time. The resulting point clouds are then synthesized off-line into maps before being uploaded to the cloud where they can be downloaded by client robots to augment navigation capabilities.
Automated Transport and Retrieval System (ATRS)
The Automatic Transport and Retrieval System (ATRS) is being developed by Freedom Sciences, LLC in conjunction with Lehigh University and Carnegie Mellon. By employing robotics and automation technologies, the core of ATRS - the "smart" wheelchair system will autonomously navigate between the driver's position and a powered lift at the rear of the vehicle.This enabling technology will allow physically challenged operators to travel independently using a standard passenger vehicle that fully conforms to NHTSA safety requirements. Furthmore, unlike alternate approaches the vehicle integration is minimally invasive, and will not affect resale value.
Our work on the ATRS program focuses on the algorithms for wheelchair position estimation and control for reliable and autonomously docking and undocking the wheelchair onto the lift platform. During the last three years, we have developed both a proof of concept system which employs a vision-based localization scheme, and the current Beta system which employs a LIDAR for estimating the wheelchair position. And the current Beta system will be improved to be 100% reliable for a planned market launch in Sep 2007.
Automated Asset Locating System (AALS)
In this project, we are merging radio-frequency identification (RFID) and mobile robotics technologies to develop an Automated Asset Locating System (AALS). Such a system will enhance inventory management by automatically updating on-hand inventory, and by locating misplaced assets in factories, warehouses, etc. – all without human interaction. For proof of concept validation, we will demonstrate a working AALS prototype that integrates computing, a LIDAR system, and RFID reader on a mobile robot base. This prototype will autonomously navigate in its environment while simultaneously locating tagged assets.
The Sick LIDAR Matlab/C++ Toolbox
The Sick LIDAR Matlab/C++ Toolbox is an open-source project released under a BSD Open-Source License that provides stable and easy-to-use C++ drivers for Sick LMS 2xx and Sick LD laser range finders. In addition to low-level drivers, the package also features an easy to use Mex interface, which allows the end-user to stream real-time range and reflectivity data directly into Matlab. This feature is especially attractive as it facilitates the rapid development of algorithms by exploiting the high-level functionality afforded by Matlab's vector-based operations. The toolbox is branched from the source code used by the Ben Franklin Racing Team during the DARPA Urban Challenge. Our entry — Little Ben — employed five Sick LMS 291 and LD-LRS LIDARs during the race, and was one of only six vehicles to successfully complete the 55 mile Urban Challenge Final Event.
The DARPA Urban Challenge
Lehigh University, the University of Pennsylvania and ATL @ Lockheed Martin have joined forces to form the Ben Franklin Racing Team, whose objective is to develop safe autonomous vehicles for urban navigation. The team is fielding an entry in the DARPA Urban Grand Challenge with Lehigh's efforts being spear-headed by our group in the VADER Lab. In this competition, autonomous vehicles must traverse a 60-mile urban course while obeying traffic laws and coping with both stationary obstacles as well as other moving traffic. Our primary vehicle is a modified hybrid Toyota Prius named Little Ben. Our focus will largely be in perception using LIDARs and stereo vision systems for road/lane segementation and obstacle detection.
Optimization Techniques for Planning and Estimation in Multi-agent Systems
We are interested in developing computationally efficient algorithms for planning and estimation, with applications to robot teams and wireless sensor networks. This work leverages recent advances in convex optimization. We attempt to extend these by establishing application specific complexity bounds and investigating distributed implementations. Topics of interest include motion planning, multi-robot localization, and target tracking.
In recent work, we investigated techniques for optimal shape changes in robot teams. Our definition of optimality is defined as either minimizing the total distance that the robots must travel or the minimizing the maximum distance that any team member must travel. Using second-order programming (SOCP) techniques, we derived optimal solutions in both SE(2) and R3. The latter also allows for complete regulation of scale.
Pennsylvania Assistive Technology Commercialization Initiative (PATCI)
PATCI is a pilot program funded through the Technology Collaborative (TTC) to encourage innovation and commercialization of Assistive Technology (AT) and Quality of Life Technology (QoLT) products. This program provides seed funding grants of up to $5,000 to student teams working on projects in AT/QoLT. Funding will be provided to student projects with commercialization potential, and which build upon technologies within s focus areas of electronics, computers, networking, and robotics. Student projects that are successful at this Phase-1 stage will be eligible to compete for Phase-2 commercialization grants of up to $30,000.
Funded student PATCI Projects include:
Hybrid Free-space Optics/Radio Frequency (FSO/RF) Networks for Mobile Robot Teams
Consider a scenario where the infrastructure of a metropolitan area network (MAN) is incapacitated by a man-made or natural disaster (e.g., an earthquake). MAN connectivity could be automatically restored by a team of autonomous mobile agents if each were equipped with a communications medium capable of patching the broken links. However, in this case the link distance and throughput requirements might be of such magnitude to render radio frequency (RF) based approaches ineffective.
We predict that such throughput intensive scenarios will lead to the emergence of hybrid FSO/RF based mobile ad-hoc network (MANET) architectures. The benefits of such a model are obvious. An FSO link in any network provides a high throughput channel over which time sensitive data can be transmitted. Many commercially available FSO devices already provide throughput rates in the Gbps range. Additionally, FSO provides constant throughput over a much longer range than RF-based channels.