Pepper Robot: What It Can Do, and Where Custom Builds Win
- David Bennett
- 2 hours ago
- 11 min read
The Pepper robot is one of the most recognizable social machines in the world. It stands at a little more than one meter tall, looks you in the eye, and responds with voice, gesture, and a tablet on its chest. As a humanoid robot for public spaces, it has set expectations for what a service robot should feel like in banks, airports, malls, and clinics.
But Pepper is also a fixed product. Its form factor, sensors, and operating system were defined years ago. Production has been paused since around 2020, and support now depends on distributors and third party teams, not on large scale manufacturing.
This article takes an unromantic, practical look at what the Pepper robot actually does well, where it struggles, and in which situations a custom build is a better long term bet. The perspective is production oriented, not speculative, and assumes you care about deployment, maintenance, and brand experience, not just novelty.
Table of Contents
What Pepper is in practical terms
Pepper is a social humanoid robot created by SoftBank Robotics (formerly Aldebaran) and produced at scale for commercial settings. It has a human inspired upper body mounted on an omni directional wheeled base, with a touchscreen in the chest, expressive arms, and a head that can turn and tilt.
Technically, you are dealing with a closed hardware and software stack:
Height around one meter twenty
Approximately twenty degrees of freedom for expressive arm and head motion
Cameras, depth sensor, microphones, and touch sensors to perceive people
A chest display as a second channel for information and input
A mobile base that moves at walking pace on flat indoor floors
From day one, Pepper was designed first as a social presence and second as a task worker. It is not a warehouse robot or a pick and place device. Its value is in greeting, guiding, and conversing, especially in front of house environments where you want something more approachable than a kiosk and more consistent than untrained staff.
Core capabilities of the Pepper platform
When you strip away the marketing language, three capability layers matter most.
1. Perception and mobility
Pepper perceives the environment at roughly human chest height. With its cameras, depth sensor, sonars, and bumper sensors, it can detect people, avoid obstacles, and navigate in open spaces.
For most deployments that means:
Detecting that someone is nearby and initiating contact
Aligning the body and head so that the robot appears to “face” the visitor
Moving through lobbies and open plan areas at safe speed
Returning to a charging point when needed
The motion model is tuned for social comfort rather than raw performance. The unit glides rather than jerks, gestures with open hands, and keeps its center of mass predictable. That makes Pepper a safe humanoid robot for crowded spaces, provided routes and no go zones are configured thoughtfully.
For a deeper treatment of sensors and perception in professional platforms, the Mimic Robotic article on smart robots and real world limits is a useful companion read.
2. Conversation and emotion recognition
Pepper combines speech recognition, text to speech, and simple emotion estimation based on facial expression and voice tone. Out of the box, it supports several major languages and can run applications built on the NAOqi framework.
In practice this enables flows such as:
Greeting visitors with context aware phrases
Asking structured questions and capturing responses on the tablet
Routing answers into backend systems or handing off to staff
Adjusting tone and body language based on basic emotional cues
As soon as you plug Pepper into a serious conversational stack with natural language understanding, it becomes much more than a menu driven kiosk. The Mimic Robotic guide to conversational AI for customer support robots illustrates how front desk flows can be designed for robots and avatars that hand over gracefully to humans.
3. App ecosystem and integration
Pepper was always envisioned as a platform, not just a single experience. There is an app model, a programming framework, and connectors that allow the robot to pull data from third party systems.
Typical integration points include:
Customer relationship management for lead capture
Queue systems and ticketing
Product information databases and content management
Analytics tools that track interactions and dwell time
This design makes the Pepper robot a flexible front end for existing digital infrastructure, as long as you invest in robust custom software rather than relying on generic demos.
Where the limitations start to show
Pepper is a fixed product that reflects the state of hardware and software design from the previous decade. For certain use cases that is completely acceptable. For others it becomes a hard constraint.
Key limitations include:
Production of new units was halted around 2020, so scaling fleets now depends on refurbished stock or remaining inventory.
The base is designed for smooth indoor floors only, which rules out outdoor guidance or rough surfaces.
The payload and reach of the arms are deliberately limited for safety, so Pepper cannot perform physical tasks beyond light gesture, simple handing over of small items, or touchscreen pointing.
Onboard compute was sized for the speech models and applications of its time; modern conversational stacks often run in the cloud rather than purely on the robot.
There are also perception limits in noisy environments. Several early retail pilots reported that background noise interfered with speech recognition and that customers preferred talking to staff when they were available.
None of this makes Pepper obsolete. It simply means you need a clear match between expectation and reality. If you treat it as a mobile, socially aware terminal that can host strong conversational content, it can still perform well. If you expect it to behave like a human staff member with full situational awareness, you will be disappointed.
Custom software on Pepper versus true custom robots
There are two different conversations that often get blurred:
Custom software running on Pepper
Custom built robots that may follow a similar body plan but with different hardware and architecture
Custom applications can take the Pepper robot much further than the stock demos. With the right team you can:
Build domain specific dialogue for banking, healthcare, education, or retail
Integrate with check in systems, loyalty platforms, and internal tools
Add multilingual flows tailored to your regions
The Mimic Robotic article on multilingual robots and global deployments outlines how language strategy and content design matter as much as hardware when you deploy in more than one country.
However, even the best software cannot change Pepper’s physical constraints. If you need a service robot that can operate on different terrain, handle specific objects, or comply with unique safety standards, at some point you are architecting around the platform rather than with it. That is the moment to consider a custom build.
Custom robots allow you to:
Choose a body form that matches your brand presence, from stylized character to near realistic attendant
Select sensors for your environment, such as additional depth cameras, thermal imaging, or industrial grade lidars
Place compute where you need it, whether on board for low latency or in local edge racks for fleet control
Design the service life, maintenance approach, and modularity of parts from the start
At that point you are much closer to the world of digital humans and virtual production, where bodies, faces, and motion are deliberately crafted rather than taken as given.
Pepper compared to bespoke service robots
To make the tradeoffs concrete, it helps to look at Pepper side by side with a custom build designed for the same class of work.
Aspect | Pepper platform | Custom service robot |
Form factor | Fixed body shape with head, arms, chest display, wheeled base | Designed from scratch for brand, environment, and tasks |
Sensing | Two cameras, depth sensor, microphones, touch sensors, basic range sensors | Sensor suite tailored to venue such as additional cameras, lidars, or beacons |
Mobility | Indoor flat floor navigation at walking pace | Options for different wheel bases, outdoor use, or constrained paths |
Conversation | Mature support from distributors and integrators, cloud based AI addons | Conversation stack designed independently of any single platform |
Brand expression | Recognizable but shared with many other organizations | Unique body, face, and motion style created for one brand |
Lifecycle | Production of new units paused, reliance on existing stock and third party support | Lifecycle defined in the contract, with agreed upgrade path and spare parts strategy |
Upfront cost | Lower initial investment per unit, especially if buying refurbished | Higher initial design and engineering cost offset by closer fit to requirements |
This is not a simple better or worse comparison. In many cases a fleet of Pepper units with strong content is absolutely the right call. In others, it becomes a stepping stone toward a dedicated humanoid robot designed around your experience. For that step you will want partners with a clear services practice around robotics and digital characters, not just hardware suppliers.
Applications
Retail and hospitality
Pepper has been used widely as a greeter and guide in shops, malls, restaurants, and hotels. It welcomes guests, offers basic information, and gathers contact details through the chest display.
Common functions include:
Welcoming visitors and answering top questions
Demonstrating loyalty programs or promotions on the tablet
Helping with wayfinding to counters, lifts, or facilities
Collecting feedback or running short surveys
In hospitality and entertainment, a custom humanoid robot might extend this further with unique choreography, costume like shells, or tight choreography with screens and light. Those projects often blend robotics with performance capture, so the same digital character appears in media, on stage, and in the lobby.
Banking and financial services
Several banks have tested Pepper as a front of house assistant that explains products, triages visitors, and handles basic FAQs so staff can focus on deeper advisory work.
For this sector, the constraints are clear. You have regulated processes, privacy sensitive data, and customers who expect precise answers. Custom builds can embed sector specific requirements into the robot’s hardware and software from the start, while also matching the institution’s visual language more closely.
Across financial, health, and cultural venues, organizations are already mapping where different types of social machines belong. The Mimic Robotic overview of industries working with service robots is a helpful reference when you plan a roadmap.
Healthcare and public services
In hospitals and clinics, Pepper acts as a calming presence in waiting areas and at information points. It can offer directions, explain procedures, and entertain children while they wait.
A custom robot for healthcare may need:
Medical grade materials and cleaning tolerance
Specific reach and motion limits around beds or equipment
Integration with hospital information systems at a deeper level
At that point a tailored build offers more long term value, with Pepper perhaps used as an early pilot to test scripts, tone, and basic flows.
Education and research
Pepper is common in labs and universities as a hands on platform for human robot interaction, pedagogy, and social robotics research.
For more advanced programs, custom robots and virtual characters often coexist. Students prototype interactions with digital humans in a game engine, then deploy the same character to a robot body and to immersive environments. This is where a connected content strategy and a shared code base begin to matter much more than any single platform.
Benefits of including Pepper in your ecosystem
Fast entry into embodied interaction
Because the Pepper robot is a known quantity, there is a global ecosystem of integrators, sample applications, and best practices.
It lets teams experiment with embodied interaction without waiting for fully bespoke hardware. That is especially valuable when you want to validate concepts such as conversational check in, social navigation, or multi party experiences with minimal hardware engineering.
Recognizable and approachable presence
Pepper’s appearance is familiar in many markets. People have seen it in media or in other venues, which lowers the barrier to interaction. The proportions are deliberately non threatening and childlike, with a stylized face that avoids the uncanny valley of more humanlike androids.
Mature support for front desk scenarios
Use cases such as reception, basic concierge tasks, and lightweight sales support have been tried and refined across many deployments. You do not need to discover every failure mode yourself. Support material, case studies, and patterns exist for queue handling, lead capture, and wayfinding.
The Mimic Robotic work on front desk conversational flows for robots documents how to design these experiences so that the robot extends your staff rather than replacing them.
Bridge to richer ecosystems
Even if your long term goal is a custom humanoid robot or a fleet of digital characters, Pepper can act as a bridge. It lets your teams get used to scripting embodied experiences, dealing with real traffic, and handling analytics and maintenance. Those lessons feed directly into larger ecosystems built around mixed fleets of robots and avatars.
When you later decide to move into a broader constellation of physical and virtual characters, an ecosystem vision like the Mimic Robotic Mimicverse of robots and digital humans gives you a way to extend beyond a single platform.
Future outlook for Pepper and custom builds
Pepper has already secured its place as an early icon of social robotics. It demonstrated that a humanoid robot in a bank or shop could feel natural rather than gimmicky when paired with the right content.
Yet the future of embodied interaction is unlikely to belong to one product. Hardware cycles are slower than software. Sensors, compute, safety requirements, and brand expectations evolve. Meanwhile, conversational AI and digital human technology move fast, independent of any specific robot.
Looking ahead:
Pepper units already in circulation will continue to operate in education, research, and select commercial sites for years, especially where there is strong local support.
New service robots are emerging with more modular hardware, more open software stacks, and tighter integration with broader ecosystems of avatars and screens.
The most resilient strategies treat robots as one manifestation of a character or assistant that also lives on web, mobile, and immersive channels.
For teams that already work with motion capture, facial rigs, and real time engines, custom builds become a natural extension. The same character that lives as a digital human in a mixed reality experience can inhabit a humanoid robot in a lobby, drawing from a shared content and AI layer.
Frequently asked questions
Is Pepper still available to buy?
New production of Pepper has been paused since around 2020. In practice, availability now depends on existing inventory and refurbished units from distributors and partners.
What can the Pepper robot realistically do in a business setting?
In a commercial environment, Pepper works best as a social assistant. It can greet visitors, answer common questions, guide people through structured flows on the tablet, collect leads, and provide simple entertainment such as short dances or quizzes. With good integration, it can also connect to booking systems or loyalty platforms.
Where does a custom build beat Pepper?
Custom robots win when the physical constraints matter. If you need specific height, reach, mobility, or sensing for your venue, or when the brand expression demands a unique body and face, a bespoke platform is more appropriate. They also make sense when you are planning a long term fleet with defined lifecycle, modularity, and upgrade path.
Can we prototype on Pepper and later move to our own robot?
Yes. Many teams prototype scripts, conversation flows, and interaction patterns on Pepper, then port the lessons to a new humanoid robot or digital human. The key is to design your content, AI stack, and analytics independently from any single piece of hardware. The Mimic Robotic blog hub collects field notes from projects that take exactly this multi platform view.
Does Pepper support multiple languages?
Pepper was designed from the start as a multilingual social robot. It supports several languages at the framework level, and integrators can add additional languages through cloud based services. For serious global deployments, it is crucial to follow robust patterns for language routing, fallback, and culturally aware content, as outlined in Mimic Robotic’s guide to multilingual robot design patterns.
Conclusion
Pepper opened the door for many organizations to think seriously about social robots. It proved that a friendly humanoid form with a screen and a decent conversational layer can lift the experience of lobbies, branches, and public venues when it is treated as part of a broader service design, not as a novelty.
At the same time, the Pepper robot is a product of its era. Its hardware is fixed, its production future is limited, and its strongest role now is as a bridge platform and a reference point. For pilots, research, and early learning in embodied interaction, Pepper remains a valuable tool.
Where you need deeper integration with your brand, your environment, and your long term roadmap, custom built service robots take the lead. They let you specify the body, the senses, and the lifecycle in the same way you would commission a high end digital human.
The most resilient strategies are not loyal to any single machine. They are loyal to the character, the experience, and the pipeline behind it. When robots, avatars, and screens all draw from the same crafted core, hardware generations can change without losing the essence of the encounter.
Comments