This case study follows the format established as part of my research into virtual pet design, which will culminate in the design of my own virtual pet.

Pleo is a robotic dinosaur ‘life form’ manufactured by Innvo Labs, first revealed in 2006 and released at retail at the end of 2007. Using a similar combination of robotics and artificial intelligence to PARO, Pleo is capable of responding to human contact and developing distinct personalities depending on its interactions with its user.

Where PARO is targeted specifically at care services, designed to aid elderly patients suffering with loneliness or the onset of dementia, Pleo uses similar technology to develop a robotic pet aimed at a commercial market.

From the PLEOworld website: ‘Every Pleo rb is autonomous. Yes, each one begins life as newly-hatched baby Camarasaurus, but that’s where predictability ends and individuality begins. Like any creature, Pleo rb feels hunger and fatigue offset by powerful urges to explore and be nurtured. It’ll graze, nap and toddle about on its own – when it feels like it! Pleo rb dinosaur can change its mind and its mood, just as you do.’

Pleo was designed by Caleb Chung, one of the co-creators of Furby, and shares much of the robotic ingenuity and appealing design that made that toy such a success.pleo-rb-autonomous-robot-life-form-3-large

Illusion of life.

Pleo uses a mixture of robotic sensors and artificial intelligence to simulate the (imagined) behaviour of a baby dinosaur. Much like Furby and PARO, the visual, mechanical and virtual design all play a part in Pleo appearing like a living creature.

This pre-release video of Pleo shows its various behavioural features:

The Pleo website outlines many of the behaviours Pleo can simulate; Pleo uses a combination of ‘born in’ characteristics, such as courage, activeness and obedience, and ‘after born’ characteristics, like hunger, mood and health, to simulate the ‘nature and nurture‘ behavioural growth of a real animal.

Pleo also ‘grows’ over time, its personality passing through four distinct life stages. It is also aware of the time of day, its behaviour changing over time, becoming active during the day and sleeping during the night.

Interaction is a big part of Pleo’s realism; it is able to respond to sight, sound and touch, much like PARO. Unlike PARO, Pleo also comes with various accessories, such as different foods that it has varying preferences for, and learning stones that enable it to ‘memorise’ different behaviours, allowing for a simulation of training to take place.

Pleo with interacts differently depending on which of its various toys you give it.

Pleo with interacts differently depending on which of its various toys you give it.

Interaction design.

Like Furby and PARO, Pleo makes use of concealed sensors and robotic parts to create convincing actions, driven by artificial intelligence. Pleo’s various ‘senses’ are outlined on the Pleo website here.

Pleo's lifelike 'senses' are provided by an advanced array of sensors.

Pleo’s lifelike ‘senses’ are provided by an advanced array of sensors.

The full set of sensors included in the original model is as follows:

  • Camera-based vision system (for light detection and navigation)
  • Two microphones, binaural hearing
  • Beat detection (allows pleo to dance and listen to music).
  • touch sensors (head, chin, shoulders, back, feet) (12 in total)
  • ARM mobile microprocessors, similar to those used in smartphones of that era, drive Pleo's intelligence.

    ARM mobile microprocessors, similar to those used in smartphones of that era, drive Pleo’s intelligence.

    Four ground foot sensors (surface detection)

  • Fourteen force-feedback sensors, one per joint
  • Orientation tilt sensor for body position
  • Infrared mouth sensor for object detection into mouth
  • Removing the outer skin gives an idea of the complex moving parts required to give Pleo believable movement.

    Removing the outer skin gives an idea of the complex moving parts required to give Pleo believable movement.

    Two-way infrared communication with other Pleos

  • Infrared detection for external objects
  • 32-bit Atmel ARM7 microprocessor (main processor for Pleo)
    • 32-bit NXP Semiconductors ARM7 sub-processor (camera system, dedicated audio input processor)
    • Four 8-bit processors (low-level motor control)

This video clip from CES2012 show a Pleo without its outer skin, and gives an idea of the complexity of the mechanical parts required to give its lifelike movements:

In 2011, Innvo Labs released the second generation of Pleo, the Pleo rb, which added more ways for users to interact with Pleo, including:

  • Recognising colors and patterns; it can also detect drop-offs so it steps back from the risk of falling.
  • Improved ‘hearing’ and language identification. Owners can name it and teach it verbal commands using “learning stones”.
  • Simulated smell and taste: it can sense what kind of “food” or “medicine” owners are feeding and then it will choose to eat or not according to its simulated needs and wants.
  • Temperature sensors in the skin, and the ability to react to heat and cold.
  • An internal clock to track the time of day.
A side-by-side comparison shows how the outer visual design of Pleo adds so much to its appeal.

A side-by-side comparison shows how the outer visual design of Pleo adds so much to its appeal.

Central to Pleo’s interaction design is the idea that its every interaction should be believable as the action of a living creature. It’s a similar design philosophy to that used in other robotic pets (Furby, PARO), where system functions are disguised as believable actions.

A feature unique to Pleo is its ability to interact with other Pleo units, as well as its user. Pleo units can have ‘conversations’ using hoots and chirps, and will approach and touch one another when left together. This video clip shows this Pleo-to-Pleo interactivity in action:

The interaction between units furthers the idea that Pleo is an independent agent, with the ability to recognise others of its kind and a set of drives that are not wholly dependant on interactions with its human user.

Social factors.

The inability of Pleo to find mainstream market success says something about the commercial viability of robotic pets, and about virtual pets in general. Pleo is a fairly advanced robotic agent, comparable to therapy robot PARO, which isn’t available at retail but costs around $2,000 per unit. It’s certainly a generational step ahead of the commercially successful Furby, a far simpler robotic pet which costs around $80 in its most recent iteration.

The result is a robotic ‘toy’ that retails for anywhere between $300 and $500. Laura DiDio, writing for Robotics Business Review in 2009, foresaw the problem facing Pleo at retail; while the unit received universal acclaim for its lifelike behaviour and believable interactions, the high price point saw it flounder at retail. This, combined with the impact of the financial crisis, saw original Pleo developer Ugobe file for bankruptcy in 2009, and Pleo itself considered to be extinct.

Each Pleo rb has a unique colour-scheme, with slight variations due to each 'skin' being handmade.

Each Pleo rb has a unique colour-scheme, with slight variations due to each ‘skin’ being handmade.

The Pleo IP was acquired by manufacturer Jetta, who founded Innvo labs to continue development of Pleo, extending the life of the original unit, and eventually rereleasing it as Pleo rb (reborn) in 2011. This rerelease didn’t alleviate the problems facing the original Pleo, however, with units retailing at a higher price point than anything else in the robotic pet marketplace, and proving difficult to find in standard retail outlets.

In the UK, Pleo rb can’t be found at common robotic pet outlets like Amazon or Toys “R” Us; in order to purchase a Pleo here one needs to search specialist robotics retailers, which list Pleo rb at anything from £300 to £500. This positioning suggest the true success of Pleo; as a niche product popular with robotics enthusiasts but with little mass market appeal.

This position is borne out by the prevalence of academic and design papers that cite Pleo. Catrinel Danauta, for example, uses Pleo as a part of a study into the ethical implications of advanced virtual pet design. Matthias Jacobsson, writing for Mobile Life Centre, has collected qualitative reports from blogs of Pleo enthusiasts, referencing the success of the unit in creating feelings of attachment and a ‘performed belief’ that sees users treating Pleo as they would a real animal.

This mixed reception to Pleo shows both the success of complex virtual pet design in creating meaningful engagements with users, and the limits of consumers’ willingness to pay for such experiences. Robotics bloggers enthuse about how truly lifelike Pleo’s behaviours are, but mainstream consumers are far more willing to pay $80 for a simplified version of these behaviour provided by Furby than they are to spend upwards of $300 on a ‘true’ robotic pet.

Share this post.