The innovative technology from Think-A-Move (TAM) called SPEAR is comprised of an in ear microphone that captures speech inside the ear canal which eliminates background noise and captures voice commands. SPEAR is designed to work with a joystick operated controlled unit and, a portable laptop. SPEAR is currently conducting field testing.
Traditionally, Unmanned Ground Vehicles (UGV) are tele-operated through a joystick-based Operator Control Unit (OCU), which tends to be big and bulky, requiring the operator to be confined to a military vehicle or some other stationary position. Moreover, joystick control of the robot requires the operator to employ both of his hands, leaving him defenseless in case of an attack (Chhatpar, Blanco, Czerniak, Hoffman, Juneja, Pruthi, Liu, Karlsen, Brown, 2009). Ttraditional ground station control systems require the operator to be secured when operating in the hostile environments typically encounter in combat scenarios. Furthermore the operator's mobility and situational awareness is degraded. In combat situations the degradation of the operator's situational awareness could become fatal. Augmenting traditional joystick actuated controlling units with SPEAR increases the operator's mobility and situational awareness.
How it works
The system design provides 30 dB of passive noise reduction, enabling it to work well in high-noise environments, where the performance of traditional speech systems, using external microphones, degrade significantly; it also utilizes a proprietary speech recognition engine to process the detected signal and additional algorithms enabling it to maintain a high level of accuracy in high noise environments… The performance of traditional speech recognition systems, which utilize an external microphone, degrades significantly when ambient noise levels increase. This is because of the decrease in signal-to-noise ratio which results from increase in ambient noise levels. Given the loud noise levels typically experienced in a military operational environment, this significantly limits the practical application of traditional speech recognition systems (Chhatpar, et al, 2009).
Once the SPEAR receives voice commands those commands are processed via a portable notebook which the operator could carry in a backpack. A wired connection carries the signal from the TAM earpiece, and ends in a standard 3.5mm audio jack that can be plugged into a computer soundcard (Brown, Blanco, Czerniak, Hoffman, Hoffman, Juneja, Ngia, Pruthi, Liu, 2010). Those commands are processed and relayed to the UGV. The UGV can execute voice commands from SPEAR and commands from the traditional joystick-based Operator Control Unit at the same time.
Challenges
Research conducted in 2010 demonstrated how important it is to tailor speech commands to the target audience. Before training the soldiers in the experiment, less than 10% of the commands the Soldiers thought should be used were the commands that were programmed into the speech-control system. Even after training and using many of the commands during a simulation task, only 34% of the Soldiers remembered the commands that the system designers programmed. Commands that were initially intuitive ("Take picture" and "Label alpha") were correctly used by 72% and 83%, respectively, of the Soldiers after training. Conversely, less intuitive phrases such as "Activate exclusion zone" were not remembered by any of the Soldiers, even after training (Redden, Carstens, Pettitt, 2010).
Another area of concern was differentiating between causal speech and speech commands. Many of the commands like "take picture" could be used in conversation with other soldiers. Having the UGV execute unintended commands during missions could be inconvenient or detrimental to the mission depending on the situation. For instance, mistakenly giving the command to "cut the wire" while explaining the details of the mission to a superior could have dire consequences during an explosive disposal operation.
Recommended Improvements
Improvements to the command data base are already in progress. Other recommended improvements are the use of a "begin command" key word and in "end command" key word. An uncommon word that would not be used in everyday speech would be most suitable. That way commands would not be activated when commands were not intended to be activated. Feedback from the SPEAR software could also ask for confirmation for sensitive commands like in the "cut the wire" example above.
The inclusion of additional user feedback references like Tactile Situational Awareness System, head tracking, gesture recognition sensors and the use of look through display and augmented reality glasses like Google Glass or Microsoft HoloLens would free up a soldiers hands for personal defense and increase the soldier's situational awareness. The soldier would be able to see through his own eyes and the camera of the UGV by utilizes a look through type display. Object recognition software could highlight and/or magnify objects of concern that the UGV sees or the soldier sees display on the look through glasses. For instance a data base of objects commonly used to hide Improvised Explosive Devices (IED) could scan the background and identify likely threats at the discretion of the operator. The operator could command the UGV camera to track his head movements during intricate operations like diffusing an IED. Tactile feedback could alert the operator when the UGV is approaching obstacles or approaching dangerous rollover angles. Gesture recognition would allow the operator to use hand signals to control the UGV when the tactical situation dictates silence. Each of these technologies are still in its infancy. Complimentary integration strategies of each control should be considered during the continued development of these technologies.
References
Brown, J., Blanco, C., Czerniak, J., Hoffman, B., Hoffman, O., Juneja, A., Ngia, L., Pruthi, T., Liu, D. (2010). Soldier experiments and assessments using SPEAR speech control system for UGVs. Paper presented at the, 7664(1) doi:10.1117/12.852507Chhatpar, S. R., Blanco, C., Czerniak, J., Hoffman, O., Juneja, A., Pruthi, T., Liu, D., Karlsen, R., Brown, J. (2009). Field experiments using SPEAR: A speech control system for UGVs. Paper presented at the, 7332(1) doi:10.1117/12.818854
Glass. Retrieved February 22, 2015 from http://www.google.com/glass/start/
Microsoft HoloLens. Retrieved February 22, 2015 from http://www.microsoft.com/microsoft-hololens/en-us
Redden, E. S., Carstens, C. B., Pettitt, R. A., & ARMY RESEARCH LAB ABERDEEN PROVING GROUND MD HUMAN RESEARCH AND ENGINEERING DIRECTORATE. (2010). Intuitive speech-based robotic control
No comments:
Post a Comment