The Potential Field Approach (PFA) [1] is a well known method in reactive robot navigation. The basic idea simply models (detected) obstacles as repulsors and goal(s) as attractor(s) depending on their proximity to the robot. The rotational of the resulting field returns the motion vectors for our mobile. PFA are intuitive, simple, smooth and, up to a point, reliable.

Despite being relatively old, this technique is still widely used in navigation. Indeed, many robotic wheelchairs use PFA to avoid collisions by removing control from the user when there is present danger in what we call "safeguard operation mode". Still, PFA per se can not be used for much more in its original formulation due to well reported problems: i) oscillations in corridor-like situations; ii) sensitivity to local minima; and iii) local traps. The following viewo shows the commented problems:

These problems can be solved by more powerful, enhanced versions of PFA, mostly based on making them less reactive. Algorithms used in assistive navigation instead of PFA include the Dynamic Window Approach and the Vector Field Histogram (VFH).


[1] Khatib, O., "Real-Time Obstacle Avoidance for Manipulators and Mobile Robots", Int. J. of Robotic Research, Vol.5, No.1 (1986), p.60

Autonomous navigation initially followed the so called sense-plan-act scheme (SPA) [Albus91], that work with a model of the environment. SPA has some well known drawbacks like strong dependence on a correct model of the environment and high latency. Reactive control architectures [Brooks86], instead, create simple behaviours by coupling sensor readings and actions and complex ones by combining several basic behaviours running concurrently. Reactive behaviours are fast, quite robust against sensor errors and noise and can easily adapt to changes in hardware or tasks. Yet, emergent behaviours are unpredictable, not necessarily efficient and prone to fall into local traps. Hybrid schemes like the well known 3T control architecture [Bonasso97] solve the aforementioned problems by combining both reactive and deliberative paradigms to achieve the best possible performance.

If one has too much time to spare in a given afternoon and some practise with Google, we can spent a few hours searching the web for companies manufacturing HCI for wheechairs and, if we are really cunning with our search, we might even get a price somewhere. This is a briefing of prices for such devices in 2009. Of course, unless you are an elf, you might need to click on the image to actually read something ... :P

In extreme, technology has recently made it possible to actually control systems with the brain. This field, widely known as BCI, is extensive, but has offered irregular results. However, there have been successful experiments in the wheelchair field lately, most related to EEG, with some work on magneto-encephalography, near-infrared spectroscopy, and functional magnetic resonance imaging as well. EEG is a favorite because it is non-invasive and fairly comfortable to use. The main problem of EEG, though, is that it provides a large amount of data with a very poor signal to noise ratio, meaning that it is actually as difficult to extract patterns from captured signals as to find the proverbial needle in a haystack. Consequently, rather than looking for very precise commands, researchers mostly quantify a reduced number of bins -sometimes labeled as mental states- to choose among a limited number of options. A typical example is trying to move a square in a screen in any of the four dominant sides -right, left, up or down-. These commands could be translated into motion directives to the wheelchair.



However, in order to fit clearly in one of these bins, the user must keep a state of continuous awareness to adequately maneuver the wheelchair. Think, for example, about juggling with several balls while following a lively conversation at the same time. Obviously, fine control here is analogous to voice-based fine control, only harder, something that could lead to excessive mental load and exhaustion. Assuming that a person can not be concentrated 24/7, some researchers found a different technique that might do the trick: rather than clustering existing signals, it is also possible to provoke a strong one and detect it. The chosen one is usually the P300 evoked potential. This natural, involuntary response of the brain to infrequent stimuli is coherent to an oddball paradigm, where a random sequence of stimuli is presented, only one of which interests the subject. Around 300ms after the target flashes, there is a positive potential peak in the EEG signal, which can be reliably detected and related to the interesting stimulus. The P300-based BCI requires almost no user training and only a few minutes to calibrate the detection algorithm param and has been successfully used to control a wheelchair.

Here I want to outline the work of the young, spanish BitBrain company, not just because they are friends, but also because they are really good at what they do.



BTW, we've gone multiligual today, but videos are pretty self-explanatory (I hope)

Wheelchair users that can not move at all are not able to use conventional HCI. Voice interfaces have been used in these situations (e.g. [Simpson,98]), so that users may actually tell the robot where they want to go. These devices, like joysticks, are commercially available and, after adjusted to a given user, fairly reliable, particularly after mobile phones have included voice recognition in their operating systems (1) Technically, they are not so different from touch screens from a qualitative point of view, as, after all, the system just receives a destination.

At some point, it was even believed that persons could simply control the chair via voice commands, in terms of left, right, slow down, stop, etc. Unfortunately, it has proven to be almost impossible to make frequent small adjustments to a wheelchair's velocity via voice [Simpson,02]. Furthermore, a failure to recognize a voice command could cause the user to be unable to travel safely and stress tends to do that, specially if users have some disability related speech difficulty. Instead, some systems decided to make adjustments on line to modify a wheelchair trajectory, previously chosen via conventional path planning instead [Rebsamen,07].

Some persons, though, have strong speech impairments and, consequently, can not use voice control as such. Voice in these cases can be complemented or replaced by other physical interfaces, controlled by head, feet, chin, shoulder switches, etc, which are fairly common, yet quite expensive in the field of assistive technologies. These interfaces are not as comfortable as the aforementioned ones but, in some cases, there may be no other choice for the user to exert some control on mobility. While it might be tempting to just intuitively choose one of these, given a particular case, it is extremely important to take into account ergonomics and medical factors. For example, some headtrackers might not be advisable for people with spinal cord injury, as they imply significant neck motion.

In more critical situations, even simpler, specifically designed interfaces can be used. After studying the needs of a patient stricked by ALS, the Telethesis project decided to use an on/off switch to choose an option in a screen that is continuously renovated. If mobility is completely out of question, more invasive interfaces are still an option. Eye tracking, for example, tries to estimate where the person is looking in order to move in that direction. Some eye tracking mechanisms are based on capturing video of the person's face to check for the position of the cornea, either with natural light or structured illumination [Li,07]. Other systems, like EagleEyes, rely on electrodes to measure the EOG, which corresponds to the angle of the eyes in the head [Yanko,98]. Electromiographic sensors use probes to capture muscular activity[Mulroy,04].

---

(1) Leading companies in the voice recognition field include Microsoft Corporation (Microsoft Voice Command), Nuance Communications (Nuance Voice Control), Vito Technology (VITO Voice2Go), Speereo Software (Speereo Voice Translator) or MyCaption for BlackBerry, to name just a few.