Automation and Robotics Strategy

Introduction

Automation and robotics have moved from isolated warehouse technologies to integrated strategic capabilities. Conveyors, sortation systems, automated storage and retrieval systems (AS/RS), shuttles, carousels, automated guided vehicles (AGVs), autonomous mobile robots (AMRs), robotic picking, and automated packing can improve throughput, accuracy, safety, space utilization, and labor resilience. They can also create substantial fixed costs, cyber-physical dependencies, integration complexity, and operational rigidity.

The central managerial question is not whether a technology is advanced. It is whether the technology fits the order profile, process constraints, service promise, labor market, facility economics, and future uncertainty of the operation. A sophisticated system can destroy value when applied to a process whose demand, product mix, or operating discipline is unsuitable. Conversely, a relatively simple conveyor or put-wall system may generate greater value than a highly autonomous solution.

Automation strategy therefore requires a systems perspective. Equipment must be evaluated as part of an order-fulfillment architecture involving inventory, information systems, labor, suppliers, maintenance, packaging, transportation cut-offs, and customers. The appropriate objective is not maximum automation but economically and operationally appropriate automation.

Learning Objectives

By the end of this chapter, students should be able to:

1. Explain the operating principles and strategic roles of major fulfillment automation technologies.
2. Match technologies to order profiles, product characteristics, and process requirements.
3. Evaluate scalability, flexibility, risk, safety, integration, and total cost of ownership.
4. Apply capacity, utilization, payback, net present value, and sensitivity concepts to automation decisions.
5. Construct and defend an automation roadmap using phased investment and governance principles.

1. Automation as a Strategic Design Decision

Automation substitutes or augments human work through mechanical, electrical, software, and analytical systems. In fulfillment, it may support one or more of five functions:

• Transport: moving inventory, totes, cartons, or orders.
• Storage: placing and retrieving inventory in dense locations.
• Accumulation: buffering work between processes.
• Identification and decision-making: scanning, dimensioning, routing, and sequencing.
• Manipulation: gripping, placing, packing, or labeling products.

A useful distinction is between fixed automation, flexible automation, and intelligent automation. Fixed automation, such as a dedicated conveyor line, performs a relatively stable sequence efficiently. Flexible automation, such as an AMR fleet, can be reconfigured through software and routing logic. Intelligent automation adds perception, learning, optimization, or exception handling, although “intelligent” does not mean autonomous under all conditions.

Automation decisions should begin with the operating model. Relevant questions include:

• What percentage of orders are single-line, multi-line, urgent, oversized, fragile, or temperature-sensitive?
• How variable are daily volume, item dimensions, and demand destinations?
• Is inventory stored in cases, totes, bins, pallets, or hanging garments?
• What service-level agreement and carrier cut-off must be achieved?
• Which activities are labor-constrained, unsafe, error-prone, or ergonomically poor?
• How frequently will the product range, facility, or business model change?

Automation should be assessed against a defined baseline, not against an abstract image of a future warehouse. The baseline includes current labor cost, throughput, accuracy, travel, space, overtime, safety incidents, quality losses, and service failures.

2. Process Fit and Order-Profile Analysis

Technology fit depends heavily on the relationship between demand characteristics and process design. Important dimensions are summarized below.

2.1 Demand and capacity

A first approximation of required average processing capacity is:

where C_req is required capacity in units per hour, V_p is planned volume in units during the planning period, and T_p is productive operating time in hours during that period. Productive time excludes planned breaks, maintenance, changeovers, and other unavailable time.

Because demand is variable, average capacity is insufficient by itself. A system may require capacity for a design peak. If V_d is daily volume, f_peak is a peak multiplier, and T_d is productive hours per day, then:

This is a planning approximation, not a guarantee of service. The peak factor should be based on historical observations, forecasts, promotions, and the desired risk tolerance. Designing for the absolute maximum can produce excessive capital cost; designing only for the mean can produce chronic overtime and missed cut-offs.

For a resource with nominal rate r and utilization u, effective capacity is approximately:

Here, u is between 0 and 1 and incorporates availability, performance losses, and quality losses. In a more explicit formulation, overall equipment effectiveness (OEE) is:

where A is availability, P is performance relative to design speed, and Q is the proportion of output meeting quality requirements. OEE is useful diagnostically, but it should not be treated as a universal measure of warehouse productivity because order fulfillment also depends on labor coordination, inventory availability, software constraints, and customer service.

High utilization is not always desirable. Queueing delay tends to rise sharply as utilization approaches 100 percent. Therefore, a system designed around continuous maximum utilization may have poor peak performance and little recovery capacity. Planned spare capacity, bypass routes, manual workstations, and redundancy are strategic choices rather than signs of inefficiency.

3. Major Technologies and Their Strategic Roles

3.1 Conveyors

Conveyors provide continuous or intermittent transport of cartons, totes, trays, pallets, or parcels. Belt, roller, chain, and sortation conveyors differ in load type, speed, accumulation behavior, and control requirements.

Their strengths include predictable routing, high throughput, reduced walking, and easy connection of receiving, storage, picking, packing, and shipping. Their weaknesses include fixed layout, significant installation cost, difficult relocation, and vulnerability to a common failure point. Conveyors are most appropriate where flows are repetitive, destinations are stable, and physical movement is a major labor burden.

Accumulation conveyors can buffer work. However, accumulation is not free: it consumes space, increases work-in-process inventory, and may conceal upstream or downstream imbalance. Conveyor design must consider carton dimensions, spacing, merge logic, accumulation pressure, emergency stops, guarding, noise, and maintenance access.

3.2 Sortation

Sortation separates items to destinations such as orders, docks, routes, stores, or pack stations. Common mechanisms include sliding-shoe, cross-belt, tilt-tray, wheel, pop-up, and shoe-wheel sorters. Selection depends on item size, speed, destination count, divert accuracy, product fragility, and required induction rate.

Sorters are valuable when many items must be routed rapidly and accurately. They can reduce manual put-wall labor, but they require reliable identification, adequate gap between items, accurate destination data, and recovery procedures for jams or unreadable labels. Sortation is not a substitute for poor upstream order consolidation; it may merely move congestion to another point.

3.3 Automated Storage and Retrieval Systems

AS/RS uses cranes, shuttles, lifts, or robotic vehicles to store and retrieve unit loads, cartons, or totes. Typical forms include pallet AS/RS, mini-load systems, tote-based systems, and cube-based robotic storage. AS/RS can provide high storage density, controlled inventory location, reduced travel, and goods-to-person picking.

The economic case is strongest where land is expensive, inventory locations are standardized, throughput is sufficiently high, and the facility will operate for many years. Risks include high capital intensity, dependence on warehouse control software, difficult retrofitting, and disruption during commissioning or major failure. Fire protection, structural loads, retrieval priorities, replenishment logic, and manual recovery must be designed from the beginning.

3.4 Shuttles

Shuttle systems use vehicles that move horizontally within storage levels, commonly with lifts transferring totes or cartons between levels. They offer modularity and high density, and additional shuttles can sometimes be added as demand grows. Their performance depends on lift capacity, aisle configuration, storage depth, replenishment policies, and the ratio of storage to retrieval activity.

Shuttles are particularly useful for tote-based goods-to-person operations with many SKUs and frequent retrievals. They are less suitable when items are highly irregular, locations change constantly, or manual access is required throughout the structure.

3.5 Carousels

Horizontal and vertical carousels rotate storage locations toward an operator. They reduce walking and can improve pick accuracy through light-directed or screen-directed confirmation. Vertical carousels can use building height efficiently and may offer secure storage for small parts.

Carousels are effective for relatively compact inventories with moderate throughput and repeated access patterns. They can become a bottleneck when a large number of simultaneous orders compete for a small number of machines. A bank of carousels, wave planning, and batching can reduce this risk, but those measures add planning complexity.

3.6 AGVs and AMRs

AGVs generally follow defined routes, markers, wires, reflectors, or mapped paths and often require controlled traffic rules. AMRs use sensors, maps, and onboard navigation to select routes and avoid obstacles dynamically. The distinction is not absolute: manufacturers use overlapping terminology, and actual capabilities must be evaluated technically.

Both technologies can transport totes, pallets, carts, or shelves. AMRs are often easier to deploy incrementally and can support goods-to-person, put-away, replenishment, and picking assistance. AGVs may provide highly repeatable movement in stable routes and industrial environments.

Fleet sizing requires more than dividing volume by nominal robot speed. Travel distance, loading and unloading time, battery charging, congestion, elevator use, traffic rules, and availability all matter. A practical model is:

where N is the required number of vehicles, D is total vehicle work in vehicle-hours during the planning period, τ is available time per vehicle in hours, u is planned utilization, and a is availability. The ceiling symbol means round upward to the next whole vehicle. The estimate is valid only when vehicle work is measured consistently and congestion does not materially change travel time. In practice, simulation or field trials should validate the result.

AMRs can increase flexibility but may introduce traffic congestion, fleet-management dependencies, charging constraints, and human-robot interaction risks. They are not automatically low-cost: the business case includes robots, racks or carts, software, network infrastructure, batteries, service contracts, and floor preparation.

3.7 Robotic picking

Robotic picking combines machine vision, product recognition, motion planning, end-effectors, and control software. It is most mature for items with predictable geometry, sufficient demand repetition, and accessible presentation. Suction, parallel-jaw, adaptive, and soft grippers each have different requirements.

The key challenge is variability. Transparent packaging, reflective surfaces, deformable bags, tightly packed bins, unknown center of gravity, and fragile products reduce successful pick probability. A useful operational metric is pick success rate:

where p_s is the proportion of successful attempts. A high success rate is necessary but not sufficient: cycle time, recovery labor, damage, replenishment, and exception handling must also be measured. Human-in-the-loop designs often outperform attempts to automate every item. Robots can handle the predictable majority while people handle exceptions.

3.8 Automated packing

Automated packing includes carton erection, item insertion, void reduction, sealing, weighing, dimensioning, labeling, and palletizing. It can reduce repetitive labor and packaging material while improving consistency. The main design issue is product and order variability. A system designed for a narrow carton range may perform poorly when order combinations change.

Packaging automation should be assessed against transportation charges, dimensional-weight rules, damage rates, material consumption, right-sizing capability, and downstream carrier requirements. A smaller carton may reduce material but increase damage if protection is inadequate. Automation must therefore optimize total delivered cost rather than packing labor alone.

4. Comparing Alternatives: A Strategic Evaluation Framework

A robust evaluation combines process fit, financial value, operational resilience, and organizational readiness. A weighted decision matrix can structure early screening, but weights must reflect strategic priorities and should not hide fatal constraints.

4.1 Total cost of ownership

Total cost of ownership (TCO) includes more than purchase price:

where C₀ is initial capital expenditure, C_t is operating, maintenance, software, training, energy, and upgrade cost in year t, i is the discount rate, n is the economic life in years, and S_n is residual or salvage value at the end. Costs and salvage value must be measured in the same currency and time basis. If risk-adjusted costs are needed, expected disruption, obsolescence, and decommissioning costs may be included explicitly rather than hidden in a contingency percentage.

Relevant cost categories include equipment, building modifications, controls, software licenses, systems integration, commissioning, spare parts, preventive maintenance, electricity, batteries, cybersecurity, insurance, labor changes, training, downtime during implementation, and eventual disposal.

4.2 Investment appraisal

Simple payback is:

where I₀ is initial investment and CF_annual is annual net cash benefit. Payback is easy to communicate but ignores the time value of money, benefits after the payback point, and residual value.

Net present value is more rigorous:

where CF_t is net cash flow in year t. An investment is financially attractive under the stated assumptions when NPV is positive relative to the chosen alternative. Cash flows should be incremental and should include taxes, working-capital effects, implementation disruption, and relevant risk assumptions where appropriate.

Worked Example: Comparing a Goods-to-Person Cell

An illustrative operation is considering a system costing $1,200,000. It expects annual labor and error savings of $420,000, additional annual maintenance and software cost of $120,000, and an annual incremental service benefit valued at $60,000. Thus annual net cash flow is:

CF = 420,000 − 120,000 + 60,000 = $360,000.

Ignoring discounting, simple payback is:

Payback = 1,200,000 / 360,000 = 3.33 years.

Assume a five-year life, no terminal value, and a discount rate of 10%. The NPV is:

NPV = −1,200,000 + 360,000 × (1 − 1.1^-5) / 0.10.

The five-year annuity factor at 10% is approximately 3.7908, so:

NPV = −1,200,000 + 360,000 × 3.7908 = $164,688 approximately.

The result is positive under these assumptions. However, the conclusion should be tested against lower volume, higher maintenance cost, implementation delay, and a shorter technology life. Financial approval should not proceed solely from the favorable base case.

5. Risk, Resilience, and Governance

Automation changes the risk profile rather than eliminating risk. A manual system may fail through absenteeism, fatigue, or inconsistent execution. An automated system may fail through software defects, sensor contamination, network outage, power loss, cyberattack, mechanical breakdown, or a common control-system failure.

A resilience assessment should identify:

• single points of failure;
• maximum tolerable downtime;
• manual bypass and degraded-mode procedures;
• critical spare parts and supplier lead times;
• backup power and network arrangements;
• recovery priorities for urgent orders;
• data backup and cybersecurity controls;
• responsibilities among the operator, integrator, and equipment supplier.

Contracts should define availability measurement, response times, acceptance testing, software escrow or access rights where relevant, data ownership, upgrade obligations, and performance remedies. “Throughput” must be precisely defined: inducted units, successfully sorted units, completed orders, or shipped orders are not equivalent measures.

Safety governance includes risk assessment, safeguarding, lockout/tagout, emergency stops, safe speeds, pedestrian segregation, training, maintenance access, and procedures for collaborative robots. Applicable legal requirements vary by jurisdiction, so organizations must consult relevant occupational safety, machinery, electrical, fire, building, and data-protection regulations. Compliance is a minimum condition, not a substitute for sound engineering.

Artificial intelligence can support vision-based picking, demand forecasting, slotting, predictive maintenance, and fleet optimization. It also introduces model drift, explainability, data-quality, privacy, and accountability concerns. AI should be deployed with measurable performance thresholds, human escalation, monitoring, version control, and a clear owner for decisions made or recommended by the system.

6. Implementation and Operating Model

Automation projects fail frequently at the interfaces between disciplines. A successful implementation normally includes the following stages:

1. Baseline: document current process times, error rates, travel, labor, constraints, and peak behavior.
2. Requirements: define product, order, service, safety, integration, and future-volume requirements.
3. Concept design: compare manual, mechanized, semi-automated, and fully automated alternatives.
4. Simulation and validation: test flow, queues, storage, replenishment, failure scenarios, and peak days.
5. Pilot: validate representative SKUs, exception rates, operator tasks, and data interfaces.
6. Detailed design and procurement: establish acceptance criteria and ownership of interfaces.
7. Commissioning: test components, integrated flows, safety functions, data, and recovery modes.
8. Ramp-up: use controlled volume increases, parallel operations, training, and daily problem solving.
9. Continuous improvement: monitor actual performance against the business case and revise operating rules.

A useful architecture separates systems by responsibility. The WMS manages inventory and order logic; a warehouse execution system may optimize work release and resource coordination; a warehouse control system manages equipment commands; programmable logic controllers control local machinery. Interfaces must be tested for timing, duplicate messages, missing confirmations, and recovery after interruption.

Operating model changes are equally important. Maintenance technicians may require electrical, mechanical, controls, and software skills. Supervisors must understand system states and exception queues. Employees should be involved in task redesign, ergonomic evaluation, and training. Automation that removes walking but creates constant exception handling may worsen work quality unless the process is redesigned holistically.

7. Illustrative Case: Northstar Health Products

Background. Northstar Health Products operates a regional distribution center serving pharmacies and clinics. It carries 18,000 SKUs, including small medical supplies, cartons, and temperature-controlled products. Annual volume is growing 12%, but demand peaks sharply before month-end. The facility experiences high picker travel, 1.8% order-line errors, overtime, and limited space for expansion.

Problem. Management initially proposes a fully automated cube-storage system with robotic picking. The proposal promises high density and reduced labor, but approximately 35% of SKUs are irregularly shaped, 20% require special handling, and several products have low demand and uncertain replenishment. A single system outage could affect the primary dispatch wave.

Analysis. The team segments the operation. Fast-moving small items have repeatable dimensions and represent 55% of order lines. Medium movers are suitable for tote-based storage but require frequent replenishment. Irregular and controlled products require human handling. The team models order peaks, replenishment travel, pick success, maintenance windows, and dispatch cut-offs. It also compares a fully automated design with a hybrid design.

Approach. Northstar selects a phased hybrid solution:

• a shuttle-based goods-to-person system for the repeatable small-item segment;
• AMRs for replenishment and movement of human-picked carts;
• a manual ergonomic zone for irregular and temperature-controlled items;
• conveyor and scan verification between picking and packing;
• semi-automated right-sized packing for standard cartons;
• a manual bypass route and duplicate critical scanning equipment.

The shuttle area is sized for current demand plus 30% growth rather than the maximum long-term forecast. The AMR fleet is introduced in stages, with congestion monitored before expansion. Robotic picking is postponed until sufficient evidence exists regarding grip success and damage rates.

Outcome. In the illustrative business case, travel in the repeatable segment falls substantially, line accuracy improves through scan confirmation, and peak overtime declines. Capital expenditure is lower than the fully automated alternative, while the manual zone preserves flexibility. The operation does not achieve maximum theoretical automation, but it improves service resilience and retains options for future expansion.

Learning points. First, segmentation is more valuable than a single technology choice. Second, postponing immature automation can be a strategic decision rather than a failure of ambition. Third, bypass capability and human exception handling should be designed before launch. Fourth, the business case must include growth, downtime, and change costs. Finally, automation should be governed by service outcomes—accurate orders shipped on time—not by robot count.

8. Contemporary Developments and Critical Perspectives

Current developments include modular micro-fulfillment, goods-to-person systems in urban facilities, robotic depalletizing, vision-guided picking, digital twins, cloud fleet management, and increasingly standardized software interfaces. These developments can lower deployment time and permit capacity to be added in modules.

However, several critical issues remain. First, vendor demonstrations may use carefully selected products and idealized flows that do not represent production variability. Second, technology life cycles may be shorter than building or equipment life cycles, creating obsolescence risk. Third, labor savings may be overstated when supervision, exception handling, maintenance, and peak staffing remain necessary. Fourth, automation can concentrate operational power in software and suppliers, increasing dependency. Fifth, social outcomes matter: job displacement, surveillance, work intensification, access to retraining, and equitable treatment should be addressed through responsible change management.

A graduate-level evaluation therefore combines engineering analysis with operations strategy, finance, organizational design, ethics, and governance. The most robust design is often a portfolio: fixed automation for stable high-volume flows, flexible automation for variable transport, human expertise for exceptions, and data systems that make the entire process visible.

Chapter Summary

Automation and robotics strategy is a process-fit and investment decision, not a technology shopping exercise. Conveyors and sorters efficiently move and route standardized flows; AS/RS, shuttles, and carousels reduce travel and increase storage density; AGVs and AMRs provide automated transport with different levels of route flexibility; robotic picking addresses manipulation but remains sensitive to product variability; and automated packing improves consistency when packaging requirements are sufficiently standardized.

Sound evaluation begins with order profiles, demand peaks, product characteristics, service requirements, and baseline performance. Capacity must account for productive time, utilization, availability, queues, and recovery needs. TCO includes capital, integration, software, energy, maintenance, labor changes, disruption, cybersecurity, and end-of-life costs. Payback is simple but incomplete; NPV provides a stronger investment perspective when cash flows, discount rates, risk, and residual value are specified clearly.

Successful systems are scalable, maintainable, safe, interoperable, and resilient. They include manual bypasses, exception processes, trained people, defined supplier responsibilities, and measurable acceptance criteria. The strategic principle is to automate appropriate work at the appropriate level, while preserving flexibility and human judgment where variability and exceptions are difficult to mechanize.

Key Terms

Automation: Use of technology to perform or support work with reduced direct human intervention.

AMR: Autonomous mobile robot that navigates dynamically using sensors, maps, and control software.

AGV: Automated guided vehicle that generally follows defined routes or guidance infrastructure.

AS/RS: Automated storage and retrieval system for mechanized placement and retrieval of inventory.

Availability: Proportion of scheduled time that equipment is operational and able to perform.

Carousel: Rotating storage system that presents locations to an operator or automated interface.

Conveyor: Mechanized equipment for transporting materials between process locations.

Goods-to-person: Fulfillment method in which inventory is brought to a stationary operator.

OEE: Product of availability, performance, and quality, used to diagnose equipment effectiveness.

Process fit: Degree to which a technology matches product, demand, flow, and service requirements.

Robotic picking: Automated manipulation and selection of items using robots, perception, and grippers.

Scalability: Ability to increase capacity or scope without disproportionate redesign or cost.

Sortation: Separation and routing of items to specified destinations.

TCO: Total cost of ownership over the life of an asset or system.

Utilization: Proportion of available capacity or time being used productively.

WCS: Warehouse control system that translates higher-level work into equipment commands.

WES: Warehouse execution system that coordinates work release, resources, and operational priorities.

WMS: Warehouse management system that manages inventory, locations, orders, and warehouse processes.

Review Questions

Short Answer Questions

1. Why is process fit more important than technological sophistication in automation strategy?
2. Distinguish between an AGV and an AMR.
3. What cost categories should be included in total cost of ownership?
4. Why can a system operating near 100% utilization perform poorly during peaks?
5. Identify two advantages and two risks of robotic picking.

Long Answer Questions

1. Compare conveyors, sortation, AS/RS, shuttles, and carousels for a high-volume e-commerce operation with substantial SKU and order variability.
2. Explain how scalability, flexibility, and resilience should influence the selection between fixed automation and AMR-based automation.
3. Discuss the limitations of simple payback and explain why NPV is generally a stronger investment appraisal method.
4. Develop an implementation roadmap for a warehouse introducing goods-to-person technology while maintaining customer service.
5. Evaluate the ethical, safety, cybersecurity, and workforce implications of a highly automated fulfillment center.

Analytical / Application Questions

1. A proposed system costs $900,000 and produces annual net cash benefits of $270,000 for five years. Calculate simple payback and explain what additional information is needed before approving the investment.
2. A facility has highly seasonal demand, irregular products, and frequent new SKU introductions. Recommend an automation portfolio and justify the balance between fixed, flexible, and manual resources.
3. Design a performance dashboard for an AMR-supported fulfillment process. Include measures for capacity, service, safety, reliability, congestion, exception handling, and financial performance.

Multiple Choice Questions with Answers

Question 1

Which condition most strongly favors fixed conveyor automation?

A. Highly unpredictable product geometry and frequent layout changes

B. Stable, high-volume flow with predictable destinations

C. Low volume with many one-off products

D. Frequent manual access to every storage location

Correct Answer: B. Stable, high-volume flow with predictable destinations

Question 2

In the capacity approximation C_eff = r × u, what does u represent?

A. Unit purchase cost

B. Number of SKUs

C. Planned utilization

D. Residual asset value

Correct Answer: C. Planned utilization

Question 3

Which technology primarily presents stored inventory to a stationary operator?

A. Goods-to-person AS/RS

B. Manual pallet jack

C. Fixed shipping dock

D. Handheld barcode scanner

Correct Answer: A. Goods-to-person AS/RS

Question 4

Why may a warehouse avoid designing automation for 100% utilization?

A. Utilization has no relationship to queues

B. High utilization can leave insufficient recovery capacity during variability

C. Low utilization always increases errors

D. Regulations prohibit utilization above 50%

Correct Answer: B. High utilization can leave insufficient recovery capacity during variability

Question 5

Which item is normally included in automation TCO?

A. Only the equipment purchase price

B. Only direct picking wages

C. Software, maintenance, energy, integration, and end-of-life costs

D. Only annual sales revenue

Correct Answer: C. Software, maintenance, energy, integration, and end-of-life costs

Question 6

A system costs $600,000 and produces annual net benefits of $150,000. What is its simple payback?

A. 2 years

B. 3 years

C. 4 years

D. 6 years

Correct Answer: C. 4 years

Question 7

What is a major challenge for robotic picking?

A. Robots cannot use sensors

B. Product variability can reduce pick success and increase exceptions

C. Robots always require a conveyor

D. Robotic picking eliminates all maintenance

Correct Answer: B. Product variability can reduce pick success and increase exceptions

Question 8

Which feature is most characteristic of an AMR compared with a traditional route-guided AGV?

A. Dynamic navigation around obstacles

B. Inability to use software

C. Requirement for a fixed rail

D. Exclusive use for pallet storage

Correct Answer: A. Dynamic navigation around obstacles

Question 9

A positive NPV generally indicates that an investment:

A. Has no operating costs

B. Recovers its investment immediately

C. Creates value relative to the stated discount rate and assumptions

D. Has zero implementation risk

Correct Answer: C. Creates value relative to the stated discount rate and assumptions

Question 10

Which implementation practice best supports resilience?

A. Eliminating all manual procedures before commissioning

B. Designing bypass routes and degraded-mode recovery procedures

C. Measuring only robot speed

D. Relying on one supplier without defined service obligations

Correct Answer: B. Designing bypass routes and degraded-mode recovery procedures